import tensorflow as tf
import matplotlib.pyplot as plt
import numpy as np
from tensorflow.keras import layers, models
from tensorflow.keras.losses import SparseCategoricalCrossentropy
from tensorflow.keras.utils import to_categorical
import numpy as np
import keras
import matplotlib.pyplot as plt
from sklearn.model_selection import KFold
from keras import models, layers
import pandas as pd
import os
from keras.models import Sequential
from keras.layers import Conv2D, MaxPooling2D, Flatten, Dense, BatchNormalization, Dropout
from keras.optimizers import Adam
from keras.preprocessing.image import ImageDataGenerator
from keras import regularizers
from keras.callbacks import EarlyStopping, ModelCheckpoint, ReduceLROnPlateau, CSVLogger, LearningRateScheduler
# from tensorflow.keras.callbacks import ReduceLROnPlateau, CSVLogger
# from tensorflow.keras.callbacks import LearningRateScheduler
import seaborn as sns
import matplotlib.pyplot as plt
import warnings
warnings.filterwarnings('ignore')

os.environ['PYTHONHASHSEED'] = '1'
keras.utils.set_random_seed(42)
tf.random.set_seed(42)
#generate random weights for reproducibility
np.random.seed(42)
import random
random.seed(42)
weights = {}
weights = {key: None for key in np.arange(0, 15)}
weights = {key: 1 for key in np.arange(0, 15)}

2023-11-26 13:03:38.404969: I tensorflow/core/platform/cpu_feature_guard.cc:182] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations.
To enable the following instructions: AVX2 AVX512F FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags.
/opt/conda/lib/python3.10/site-packages/tensorflow_io/python/ops/__init__.py:98: UserWarning: unable to load libtensorflow_io_plugins.so: unable to open file: libtensorflow_io_plugins.so, from paths: ['/opt/conda/lib/python3.10/site-packages/tensorflow_io/python/ops/libtensorflow_io_plugins.so']
caused by: ['/opt/conda/lib/python3.10/site-packages/tensorflow_io/python/ops/libtensorflow_io_plugins.so: undefined symbol: _ZN3tsl6StatusC1EN10tensorflow5error4CodeESt17basic_string_viewIcSt11char_traitsIcEENS_14SourceLocationE']
  warnings.warn(f"unable to load libtensorflow_io_plugins.so: {e}")
/opt/conda/lib/python3.10/site-packages/tensorflow_io/python/ops/__init__.py:104: UserWarning: file system plugins are not loaded: unable to open file: libtensorflow_io.so, from paths: ['/opt/conda/lib/python3.10/site-packages/tensorflow_io/python/ops/libtensorflow_io.so']
caused by: ['/opt/conda/lib/python3.10/site-packages/tensorflow_io/python/ops/libtensorflow_io.so: undefined symbol: _ZTVN10tensorflow13GcsFileSystemE']
  warnings.warn(f"file system plugins are not loaded: {e}")

Background Research

train_data = tf.keras.utils.image_dataset_from_directory(
    'test/train/',
    labels='inferred',
    label_mode='int',
    class_names=None,
    color_mode='grayscale',
    batch_size=32,
    image_size=(224, 224),
    shuffle=True,
    seed=42,
    validation_split=None,
    subset=None,
    interpolation='bilinear',
    crop_to_aspect_ratio=False,
)

Found 9028 files belonging to 15 classes.

2023-11-26 13:03:47.092469: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2023-11-26 13:03:47.123136: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2023-11-26 13:03:47.123425: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2023-11-26 13:03:47.125113: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2023-11-26 13:03:47.125377: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2023-11-26 13:03:47.125556: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2023-11-26 13:03:47.921463: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2023-11-26 13:03:47.921713: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2023-11-26 13:03:47.921897: I tensorflow/compiler/xla/stream_executor/cuda/cuda_gpu_executor.cc:996] successful NUMA node read from SysFS had negative value (-1), but there must be at least one NUMA node, so returning NUMA node zero. See more at https://github.com/torvalds/linux/blob/v6.0/Documentation/ABI/testing/sysfs-bus-pci#L344-L355
2023-11-26 13:03:47.922061: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1635] Created device /job:localhost/replica:0/task:0/device:GPU:0 with 14587 MB memory:  -> device: 0, name: Tesla V100-SXM2-16GB, pci bus id: 0000:00:04.0, compute capability: 7.0

for batch in train_data.take(10):
    print(batch[1][0])

tf.Tensor(0, shape=(), dtype=int32)
tf.Tensor(11, shape=(), dtype=int32)
tf.Tensor(10, shape=(), dtype=int32)
tf.Tensor(6, shape=(), dtype=int32)
tf.Tensor(3, shape=(), dtype=int32)
tf.Tensor(9, shape=(), dtype=int32)
tf.Tensor(3, shape=(), dtype=int32)
tf.Tensor(14, shape=(), dtype=int32)
tf.Tensor(14, shape=(), dtype=int32)
tf.Tensor(5, shape=(), dtype=int32)

2023-11-26 13:03:48.168431: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'Placeholder/_0' with dtype string and shape [9028]
	 [[{{node Placeholder/_0}}]]
2023-11-26 13:03:48.168971: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'Placeholder/_4' with dtype int32 and shape [9028]
	 [[{{node Placeholder/_4}}]]

validation_data = tf.keras.utils.image_dataset_from_directory(
    'test/validation/',
    labels='inferred',
    label_mode='int',
    class_names=None,
    color_mode='grayscale',
    batch_size=32,
    image_size=(224, 224),
    shuffle=True,
    seed=42,
    validation_split=None,
    subset=None,
    interpolation='bilinear',
    crop_to_aspect_ratio=False,
)

Found 3000 files belonging to 15 classes.

test_data = tf.keras.utils.image_dataset_from_directory(
    'test/test/',
    labels='inferred',
    label_mode='int',
    class_names=None,
    color_mode='grayscale',
    batch_size=32,
    image_size=(224, 224),
    shuffle=True,
    seed=42,
    validation_split=None,
    subset=None,
    interpolation='bilinear',
    crop_to_aspect_ratio=False,
)

Found 3000 files belonging to 15 classes.

Exploratory Data Analysis

Exploratory Data Analysis (EDA) is a critical step in understanding and preparing data for various purposes, including machine learning and data-driven decision-making. We can perform EDA to further our understanding of the data, and to make better decisions on how we can process it to make it easier for our model to understand.

#sample to check if the data is loaded correctly

for batch in train_data.take(1):
    plt.figure()
    plt.imshow(batch[0][0]) 
    plt.title(f"Label: {batch[1][0]}")
    plt.show()


for batch in test_data.take(1):
    plt.figure()
    plt.imshow(batch[0][0], cmap='gray') 
    plt.title(f"Label: {batch[1][0]}")
    plt.show()


for batch in validation_data.take(1):
    plt.figure()
    plt.imshow(batch[0][0], cmap='gray') 
    plt.title(f"Label: {batch[1][0]}")
    plt.show()

2023-11-26 13:03:48.907782: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'Placeholder/_4' with dtype int32 and shape [9028]
	 [[{{node Placeholder/_4}}]]
2023-11-26 13:03:48.908404: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'Placeholder/_0' with dtype string and shape [9028]
	 [[{{node Placeholder/_0}}]]

2023-11-26 13:03:49.301475: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'Placeholder/_0' with dtype string and shape [3000]
	 [[{{node Placeholder/_0}}]]
2023-11-26 13:03:49.301947: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'Placeholder/_0' with dtype string and shape [3000]
	 [[{{node Placeholder/_0}}]]

2023-11-26 13:03:49.598494: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'Placeholder/_4' with dtype int32 and shape [3000]
	 [[{{node Placeholder/_4}}]]
2023-11-26 13:03:49.599162: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'Placeholder/_0' with dtype string and shape [3000]
	 [[{{node Placeholder/_0}}]]

import matplotlib.pyplot as plt

def plot_samples_by_label(data):
    # Create a dictionary to store one sample per label
    samples_by_label = {}
    # Iterate through the training data
    for batch in data:
        image, label = batch[0][0], batch[1][0].numpy()
        
        # Check if a sample for this label already exists
        if label not in samples_by_label:
            samples_by_label[label] = image
            
        # Check if we have samples for all labels
        if len(samples_by_label) == len(set(labels)):
            break
    #use subplot to plot a 5x3 grid of images
    fig, axes = plt.subplots(nrows=3, ncols=5, figsize=(15,8))
    # Loop through each label and subplot
    for label, image in samples_by_label.items():
        row = int(label // 5)
        col = label % 5
        ax = axes[row, col]
        ax.imshow(image, cmap='gray')
        ax.axis('off')
        ax.set_title(f"Label: {label}")


labels = [batch[1][0].numpy() for batch in train_data]
# Call the function with your training data
plot_samples_by_label(train_data)

2023-11-26 13:03:49.910977: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'Placeholder/_0' with dtype string and shape [9028]
	 [[{{node Placeholder/_0}}]]
2023-11-26 13:03:49.911580: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'Placeholder/_0' with dtype string and shape [9028]
	 [[{{node Placeholder/_0}}]]

class_counts = {}


for images, labels in train_data:
    for label in labels.numpy():
        class_name = train_data.class_names[label]
        if class_name in class_counts:
            class_counts[class_name] += 1
        else:
            class_counts[class_name] = 1


train_class=pd.DataFrame(list(class_counts.items()),columns=['Class','count'])
train_class=train_class.sort_values(by=['count'],ascending=False)
train_class.plot(kind='bar',figsize=(10,5),title="Number of images in each class",x='Class',y='count')

<Axes: title={'center': 'Number of images in each class'}, xlabel='Class'>

Interpreting the Previous Output

Based on the output of the previous cell, we can see that the dataset is not evenly distributed across all labels. This can be problematic for our model, as it may not be able to learn from the data effectively if it is biased towards certain labels.

Shifting Weights in the Model

One way to address this issue is to shift the weights in the model to account for the uneven distribution of labels. This can be done by assigning higher weights to underrepresented labels and lower weights to overrepresented labels.

For example, we can use the class_weight parameter in Keras to assign weights to each class based on their frequency in the dataset. This will ensure that the model pays more attention to underrepresented classes during training.

Feature engineering or data augmentation

# Convert datasets to NumPy arrays
train_images, train_labels = zip(*[(image, label) for image, label in train_data.as_numpy_iterator()])
validation_images, validation_labels = zip(*[(image, label) for image, label in validation_data.as_numpy_iterator()])
test_images,test_labels= zip(*[(image, label) for image, label in test_data.as_numpy_iterator()])
#train set
X_train = np.concatenate(train_images, axis=0)
y_train = np.concatenate(train_labels, axis=0)

#val
X_val = np.concatenate(validation_images, axis=0) 
y_val = np.concatenate(validation_labels, axis=0)

#test
X_test = np.concatenate(test_images, axis=0)
y_test = np.concatenate(test_labels, axis=0)

2023-11-26 13:03:56.202024: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'Placeholder/_4' with dtype int32 and shape [3000]
	 [[{{node Placeholder/_4}}]]
2023-11-26 13:03:56.202344: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'Placeholder/_4' with dtype int32 and shape [3000]
	 [[{{node Placeholder/_4}}]]
2023-11-26 13:03:56.604515: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'Placeholder/_4' with dtype int32 and shape [3000]
	 [[{{node Placeholder/_4}}]]
2023-11-26 13:03:56.604996: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'Placeholder/_4' with dtype int32 and shape [3000]
	 [[{{node Placeholder/_4}}]]

# Resize images to (31, 31) using TensorFlow
X_train_resized = tf.image.resize(X_train, (31, 31))
X_val_resized = tf.image.resize(X_val, (31, 31))

# Reshape and normalize
X_train31 = X_train_resized.numpy().reshape(X_train_resized.shape[0], 31, 31, 1).astype('float32') / 255
X_val31 = X_val_resized.numpy().reshape(X_val_resized.shape[0], 31, 31, 1).astype('float32') / 255


validation_labels = to_categorical(y_val)
train_labels = to_categorical(y_train)

Modelling

Base Model

We can first try fitting the data in an extremely simple model to get a baseline understanding of the problem. This initial model serves as a reference point for evaluating the performance of more complex models. The base model will be straightforward and quick to implement, which provides a easy way capture the underlying patterns in the data.

Key Characteristics of the Base Model:

• Simplicity: The base model is intentionally kept simple to establish a basic understanding of the data and the problem at hand.

• Quick Implementation: It is designed for rapid implementation, allowing for a fast assessment of initial performance.

• Benchmark: The base model serves as a benchmark against which more sophisticated models can be compared. Any subsequent model should outperform this baseline to be considered effective.

Evaluation Metric:

The evaluation metric we will be using will be val accuracy as well as val loss, this will allow us to rougly understand the model's accuracy (val accuracy) as well as its rate of error (val loss)

import matplotlib.pyplot as plt

def plot_history(history):
    fig, axes = plt.subplots(nrows=1, ncols=2, figsize=(15,8))
    # Plot training & validation accuracy values
    ax = axes[1]
    ax.plot(history.history['accuracy'])
    ax.plot(history.history['val_accuracy'])
    ax.set_title('Model accuracy')
    ax.set_ylabel('Accuracy')
    ax.set_xlabel('Epoch')
    ax.legend(['Train', 'Validation'], loc='upper left')

    # Plot training & validation loss values
    ax = axes[0]
    ax.plot(history.history['loss'])
    ax.plot(history.history['val_loss'])
    ax.set_title('Model loss')
    ax.set_ylabel('Loss')
    ax.set_xlabel('Epoch')
    ax.legend(['Train', 'Validation'], loc='upper left')

df31 = pd.DataFrame(columns = ['Model Name', 'Train Accuracy', 'Validation Accuracy', 'Train Loss', 'Validation Loss', 'History'])


model_31 = models.Sequential([
    layers.Conv2D(32, (3, 3), activation='relu', input_shape=(31, 31, 1)),
    layers.MaxPooling2D((2, 2)),
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.MaxPooling2D((2, 2)),
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.Flatten(),
    layers.Dense(64, activation='relu'),
    layers.Dense(15, activation='softmax')
])

# Compile the model
model_31.compile(optimizer='adam',
              loss= 'categorical_crossentropy',
              metrics=['accuracy'])

# Train the model
history = model_31.fit(X_train31, train_labels, validation_data=(X_val31, validation_labels),
 epochs=10, batch_size=32, verbose=2, class_weight = weights)
scores = model_31.evaluate(X_val31, validation_labels, verbose=0)
print("Baseline Error: %.2f%%" % (100-scores[1]*100))
plot_history(history)
df31.loc[len(df31)] = ['Base 31 Model', history.history['accuracy'][-1], history.history['val_accuracy'][-1], history.history['loss'][-1], history.history['val_loss'][-1], history]

Epoch 1/10

2023-11-26 13:04:02.043106: I tensorflow/compiler/xla/stream_executor/cuda/cuda_dnn.cc:424] Loaded cuDNN version 8900
2023-11-26 13:04:02.902544: I tensorflow/compiler/xla/service/service.cc:169] XLA service 0x7f53234fb0b0 initialized for platform CUDA (this does not guarantee that XLA will be used). Devices:
2023-11-26 13:04:02.902589: I tensorflow/compiler/xla/service/service.cc:177]   StreamExecutor device (0): Tesla V100-SXM2-16GB, Compute Capability 7.0
2023-11-26 13:04:02.910062: I tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:269] disabling MLIR crash reproducer, set env var `MLIR_CRASH_REPRODUCER_DIRECTORY` to enable.
2023-11-26 13:04:03.072317: I ./tensorflow/compiler/jit/device_compiler.h:180] Compiled cluster using XLA!  This line is logged at most once for the lifetime of the process.

283/283 - 6s - loss: 2.3945 - accuracy: 0.1946 - val_loss: 2.2241 - val_accuracy: 0.3070 - 6s/epoch - 20ms/step
Epoch 2/10
283/283 - 1s - loss: 1.8913 - accuracy: 0.3858 - val_loss: 1.9207 - val_accuracy: 0.3587 - 1s/epoch - 4ms/step
Epoch 3/10
283/283 - 1s - loss: 1.6249 - accuracy: 0.4708 - val_loss: 1.7104 - val_accuracy: 0.4443 - 1s/epoch - 4ms/step
Epoch 4/10
283/283 - 1s - loss: 1.3880 - accuracy: 0.5430 - val_loss: 1.4243 - val_accuracy: 0.5307 - 1s/epoch - 4ms/step
Epoch 5/10
283/283 - 1s - loss: 1.2080 - accuracy: 0.6014 - val_loss: 1.2698 - val_accuracy: 0.5850 - 1s/epoch - 4ms/step
Epoch 6/10
283/283 - 1s - loss: 1.0624 - accuracy: 0.6492 - val_loss: 1.0915 - val_accuracy: 0.6340 - 1s/epoch - 4ms/step
Epoch 7/10
283/283 - 1s - loss: 0.9394 - accuracy: 0.6905 - val_loss: 1.1373 - val_accuracy: 0.6290 - 1s/epoch - 4ms/step
Epoch 8/10
283/283 - 1s - loss: 0.8452 - accuracy: 0.7226 - val_loss: 1.1535 - val_accuracy: 0.6383 - 1s/epoch - 4ms/step
Epoch 9/10
283/283 - 1s - loss: 0.7634 - accuracy: 0.7529 - val_loss: 1.0070 - val_accuracy: 0.6637 - 1s/epoch - 4ms/step
Epoch 10/10
283/283 - 1s - loss: 0.6793 - accuracy: 0.7801 - val_loss: 1.5060 - val_accuracy: 0.5377 - 1s/epoch - 4ms/step
Baseline Error: 46.23%

model_31.save('Base31Model.h5')

X_train_resized = tf.image.resize(X_train, (128, 128))
X_val_resized = tf.image.resize(X_val, (128, 128))

# Reshape and normalize
X_train128 = X_train_resized.numpy().reshape(X_train_resized.shape[0], 128, 128, 1).astype('float32') / 255
X_val128 = X_val_resized.numpy().reshape(X_val_resized.shape[0], 128, 128, 1).astype('float32') / 255

df128 = pd.DataFrame(columns = ['Model', 'Train Accuracy', 'Validation Accuracy', 'Train Loss', 'Validation Loss', 'History'])

model_128 = models.Sequential([
    layers.Conv2D(32, (3, 3), activation='relu', input_shape=(128, 128, 1)),
    layers.MaxPooling2D((2, 2)),
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.MaxPooling2D((2, 2)),
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.Flatten(),
    layers.Dense(64, activation='relu'),
    layers.Dense(15, activation='softmax')
])

# Compile the model
model_128.compile(optimizer='adam',
              loss= 'categorical_crossentropy',
              metrics=['accuracy'])

# Train the model
history = model_128.fit(X_train128, train_labels, validation_data=(X_val128, validation_labels),
 epochs=10, batch_size=32, verbose=2, class_weight = weights)
scores = model_128.evaluate(X_val128, validation_labels, verbose=0)
print("Baseline Error: %.2f%%" % (100-scores[1]*100))
plot_history(history)
df128.loc[len(df128)] = ['Base 128 Model', history.history['accuracy'][-1], history.history['val_accuracy'][-1], history.history['loss'][-1], history.history['val_loss'][-1], history]

Epoch 1/10
283/283 - 5s - loss: 1.9846 - accuracy: 0.3630 - val_loss: 1.7195 - val_accuracy: 0.4307 - 5s/epoch - 16ms/step
Epoch 2/10
283/283 - 2s - loss: 1.1167 - accuracy: 0.6564 - val_loss: 1.0886 - val_accuracy: 0.6613 - 2s/epoch - 7ms/step
Epoch 3/10
283/283 - 2s - loss: 0.6139 - accuracy: 0.8137 - val_loss: 0.7902 - val_accuracy: 0.7620 - 2s/epoch - 7ms/step
Epoch 4/10
283/283 - 2s - loss: 0.3018 - accuracy: 0.9094 - val_loss: 0.7090 - val_accuracy: 0.7850 - 2s/epoch - 7ms/step
Epoch 5/10
283/283 - 2s - loss: 0.1490 - accuracy: 0.9555 - val_loss: 0.9785 - val_accuracy: 0.7557 - 2s/epoch - 7ms/step
Epoch 6/10
283/283 - 2s - loss: 0.1152 - accuracy: 0.9657 - val_loss: 0.8711 - val_accuracy: 0.8013 - 2s/epoch - 7ms/step
Epoch 7/10
283/283 - 2s - loss: 0.0481 - accuracy: 0.9869 - val_loss: 1.0326 - val_accuracy: 0.7907 - 2s/epoch - 7ms/step
Epoch 8/10
283/283 - 2s - loss: 0.0721 - accuracy: 0.9802 - val_loss: 0.9543 - val_accuracy: 0.8033 - 2s/epoch - 7ms/step
Epoch 9/10
283/283 - 2s - loss: 0.0277 - accuracy: 0.9917 - val_loss: 1.1317 - val_accuracy: 0.7967 - 2s/epoch - 7ms/step
Epoch 10/10
283/283 - 2s - loss: 0.0453 - accuracy: 0.9855 - val_loss: 1.2959 - val_accuracy: 0.7693 - 2s/epoch - 7ms/step
Baseline Error: 23.07%

model_128.save('Base128Model.h5')

Base Model Evaluation

As we can see, the base model has achieved a decent accuracy of around 75% for both after 10 epochs. This level of accuracy provides valuable insights into the nature of the dataset and indicates that the dataset is suitable for modeling.

Accuracy Interpretation:

• Decent Performance: An accuracy of 75% suggests that the base model is able to correctly predict the target variable for 75% of the instances in the dataset. This indicates a solid performance, but there may still be room for improvement.

• Model Feasibility: The achieved accuracy reaffirms that the dataset contains discernible patterns that the model is capturing. The feasibility of modeling is supported by the ability to achieve a reasonably high accuracy.

• Baseline for Comparison: The 75% accuracy serves as a baseline for evaluating the performance of more complex models. Any subsequent model should aim to surpass this accuracy to demonstrate significant improvement.

Loss Interpretation:

128x128 Graph:

• Training Progress: The loss decreases over the 10 epochs, indicating effective learning from the training data. However we should also note the increase in loss after the 5th epoch, as this is the base model there is really no need to take precaution for this specific model, however we should keep in mind that when we are trying other actual model structures, we can use a callback to prevent the model from overfitting.

• Validation Loss Interpretation: The validation loss also decreases, suggesting that the model generalizes well to unseen data.

31x31 Graph:

• Training Progress: Similar to the other graph, the val-loss decreases after 10 epochs, showing effective learning from the training data.

• Validation Loss Interpretation: The validation loss decreases, indicating good generalization to unseen data.

Comparison:

• Accuracy Comparison: Both 128x128 and 31x31 models achieve around 75% accuracy, suggesting that the model is performing consistently across different resolutions.

• Loss Comparison: The loss graphs for both resolutions show a consistent decrease, indicating effective learning. The smaller graph (31x31) might have slightly more fluctuations, but the overall trend is still downward.

In summary, the base model demonstrates promising performance, however some overfitting can be seen and further optimization and exploration can likely yield even better results.

Model Improvements

1. Advanced Architectures:
   • Description: Consider more advanced neural network architectures such as ResNet, Inception, or EfficientNet, depending on the nature of the problem. These architectures often capture intricate patterns more effectively.
   • Benefits: Enhanced representational capacity can lead to better model performance.

1. Data Augmentation:

   • Description: Data augmentation involves applying random transformations to the training data, such as rotation, scaling, and flipping. This increases the diversity of the training set, helping the model generalize better to different variations of the input data.
   • Benefits: Mitigates overfitting by exposing the model to a wider range of variations in the data.

2. Batch Normalization:

   • Description: Batch normalization normalizes the input of each layer in a neural network to have zero mean and unit variance. This can accelerate training, improve model stability, and reduce sensitivity to weight initialization.
   • Benefits: Enhances convergence speed and can make the model less dependent on the choice of hyperparameters.

3. Learning Rate Scheduling:

   • Description: Adjusting the learning rate during training can be beneficial. For example, starting with a higher learning rate and gradually reducing it as training progresses can help the model converge more effectively.
   • Benefits: Can improve convergence, prevent overshooting, and result in a more stable training process.

4. Regularization Techniques:

   • Description: Techniques like dropout, L1, and L2 regularization can be applied to prevent overfitting. Dropout randomly drops units during training, while L1 and L2 regularization penalize large weights.
   • Benefits: Helps the model generalize better to new data and reduces the risk of overfitting.

5. Early Stopping:

   • Description: Monitor the validation loss during training and stop training when the loss stops improving. This prevents overfitting and saves computational resources.
   • Benefits: Guards against overfitting and ensures the model is not trained for too many epochs.
6. Adjusting Weights:
   • Description: We can adjust the weights in order to account for the imbalanced distribution of the dataset.
   • Benefits: Helps address the issue of imbalanced classes by assigning higher weights to under-represented classes during training. This ensures that the model pays more attention to minority classes, improving overall predictive performance and reducing bias towards the majority class.

3. Model Structures/Advanced Architecture

In our quest to enhance the performance of our neural network, we'll explore two well-established architectures designed by experts: AlexNet and LeNet.

1. AlexNet:

• Description:
  • Input Layer:
    • Shape: 227x227x3 (color images with RGB channels).
  • Convolutional Layer 1:
    • Filters: 96
    • Kernel Size: 11x11
    • Activation Function: ReLU
    • Max Pooling: 3x3, stride 2
  • Convolutional Layer 2:
    • Filters: 256
    • Kernel Size: 5x5
    • Activation Function: ReLU
    • Max Pooling: 3x3, stride 2
  • Convolutional Layer 3:
    • Filters: 384
    • Kernel Size: 3x3
    • Activation Function: ReLU
  • Convolutional Layer 4:
    • Filters: 384
    • Kernel Size: 3x3
    • Activation Function: ReLU
  • Convolutional Layer 5:
    • Filters: 256
    • Kernel Size: 3x3
    • Activation Function: ReLU
    • Max Pooling: 3x3, stride 2
  • Fully Connected Layer 1:
    • Neurons: 4096
    • Activation Function: ReLU
  • Fully Connected Layer 2:
    • Neurons: 4096
    • Activation Function: ReLU
  • Output Layer:
    • Neurons: Number of classes in the task
    • Activation Function: Typically softmax for classification tasks.

2. LeNet:

• Description:
  • Input Layer:
    • Shape: 32x32 (for grayscale images) or 32x32x3 (for color images with RGB channels).
  • Convolutional Layer 1:
    • Filters: 6
    • Kernel Size: 5x5
    • Activation Function: Tanh
    • Max Pooling: 2x2, stride 2
  • Convolutional Layer 2:
    • Filters: 16
    • Kernel Size: 5x5
    • Activation Function: Tanh
    • Max Pooling: 2x2, stride 2
  • Fully Connected Layer 1:
    • Neurons: 120
    • Activation Function: Tanh
  • Fully Connected Layer 2:
    • Neurons: 84
    • Activation Function: Tanh
  • Output Layer:
    • Neurons: Number of classes in the task
    • Activation Function: Typically softmax for classification tasks.

The input layer will be scaled to fit the images we are using (31x31 and 128x128)

def AlexNet(input_shape, num_classes=15):
    model = models.Sequential()
    # Convolutional Layer 1
    model.add(layers.Conv2D(96, (11, 11), strides=(4, 4), activation='relu', input_shape=(input_shape, input_shape, 1)))
    model.add(layers.MaxPooling2D((2, 2), strides=(2, 2)))

    # Convolutional Layer 2
    model.add(layers.Conv2D(256, (5, 5), padding='same', activation='relu'))

    # Adjust the pooling parameters or remove pooling if necessary
    # model.add(layers.MaxPooling2D((2, 2), strides=(2, 2)))

    # Convolutional Layer 3
    model.add(layers.Conv2D(384, (3, 3), padding='same', activation='relu'))

    # Convolutional Layer 4
    model.add(layers.Conv2D(384, (3, 3), padding='same', activation='relu'))

    # Convolutional Layer 5
    model.add(layers.Conv2D(256, (3, 3), padding='same', activation='relu'))
    model.add(layers.MaxPooling2D((3, 3), strides=(2, 2)))

    # Flatten the output before the fully connected layers
    model.add(layers.Flatten())

    # Fully Connected Layer 1
    model.add(layers.Dense(4096, activation='relu'))

    # Fully Connected Layer 2
    model.add(layers.Dense(4096, activation='relu'))

    # Output Layer
    model.add(layers.Dense(num_classes, activation='softmax'))

    return model

def Lenet(input_shape, num_classes=15):
    model = models.Sequential()

    # Convolutional Layer 1
    model.add(layers.Conv2D(6, (5, 5), activation='tanh', input_shape=(input_shape, input_shape,1)))
    model.add(layers.MaxPooling2D((2, 2), strides=(2, 2)))

    # Convolutional Layer 2
    model.add(layers.Conv2D(16, (5, 5), activation='tanh'))
    model.add(layers.MaxPooling2D((2, 2), strides=(2, 2)))

    # Flatten the output before the fully connected layers
    model.add(layers.Flatten())

    # Fully Connected Layer 1
    model.add(layers.Dense(120, activation='tanh'))

    # Fully Connected Layer 2
    model.add(layers.Dense(84, activation='tanh'))

    # Output Layer
    model.add(layers.Dense(num_classes, activation='softmax'))

    return model

# 31x31 LeNet

model = Lenet(31)
model.compile(optimizer='adam',
              loss= 'categorical_crossentropy',
              metrics=['accuracy'])
history = model.fit(X_train31, train_labels, validation_data=(X_val31, validation_labels),
 epochs=10, batch_size=32, verbose=2, class_weight = weights)
scores = model.evaluate(X_val31, validation_labels, verbose=0)
print("Baseline Error: %.2f%%" % (100-scores[1]*100))

plot_history(history)
df31.loc[len(df31)] = ['LeNet', history.history['accuracy'][-1], history.history['val_accuracy'][-1], history.history['loss'][-1], history.history['val_loss'][-1], history]

Epoch 1/10
283/283 - 4s - loss: 2.3475 - accuracy: 0.2483 - val_loss: 2.2139 - val_accuracy: 0.3003 - 4s/epoch - 13ms/step
Epoch 2/10
283/283 - 1s - loss: 1.9472 - accuracy: 0.3871 - val_loss: 1.9224 - val_accuracy: 0.3863 - 1s/epoch - 4ms/step
Epoch 3/10
283/283 - 1s - loss: 1.6955 - accuracy: 0.4649 - val_loss: 1.8224 - val_accuracy: 0.4173 - 1s/epoch - 4ms/step
Epoch 4/10
283/283 - 1s - loss: 1.5319 - accuracy: 0.5200 - val_loss: 1.6240 - val_accuracy: 0.4850 - 1s/epoch - 4ms/step
Epoch 5/10
283/283 - 1s - loss: 1.3913 - accuracy: 0.5651 - val_loss: 1.5202 - val_accuracy: 0.5277 - 1s/epoch - 4ms/step
Epoch 6/10
283/283 - 1s - loss: 1.2777 - accuracy: 0.6025 - val_loss: 1.4312 - val_accuracy: 0.5390 - 1s/epoch - 4ms/step
Epoch 7/10
283/283 - 1s - loss: 1.1565 - accuracy: 0.6382 - val_loss: 1.3474 - val_accuracy: 0.5633 - 1s/epoch - 4ms/step
Epoch 8/10
283/283 - 1s - loss: 1.0620 - accuracy: 0.6674 - val_loss: 1.3021 - val_accuracy: 0.5867 - 1s/epoch - 4ms/step
Epoch 9/10
283/283 - 1s - loss: 0.9834 - accuracy: 0.6953 - val_loss: 1.2106 - val_accuracy: 0.6163 - 1s/epoch - 4ms/step
Epoch 10/10
283/283 - 1s - loss: 0.8788 - accuracy: 0.7286 - val_loss: 1.3322 - val_accuracy: 0.5760 - 1s/epoch - 4ms/step
Baseline Error: 42.40%

# 31x31 alexnet 

model = AlexNet(31)
model.compile(optimizer='adam',
              loss= 'categorical_crossentropy',
              metrics=['accuracy'])
history = model.fit(X_train31, train_labels, validation_data=(X_val31, validation_labels),
                    epochs=10, batch_size=32, verbose=2, class_weight = weights)


scores = model.evaluate(X_val31, validation_labels, verbose=0)
print("Baseline Error: %.2f%%" % (100-scores[1]*100))
plot_history(history)
df31.loc[len(df31)] = ['AlexNet', history.history['accuracy'][-1], history.history['val_accuracy'][-1], history.history['loss'][-1], history.history['val_loss'][-1], history]

Epoch 1/10
283/283 - 6s - loss: 2.6459 - accuracy: 0.0984 - val_loss: 2.7833 - val_accuracy: 0.0667 - 6s/epoch - 20ms/step
Epoch 2/10
283/283 - 2s - loss: 2.6401 - accuracy: 0.1031 - val_loss: 2.7889 - val_accuracy: 0.0667 - 2s/epoch - 6ms/step
Epoch 3/10
283/283 - 2s - loss: 2.6384 - accuracy: 0.1037 - val_loss: 2.7829 - val_accuracy: 0.0667 - 2s/epoch - 6ms/step
Epoch 4/10
283/283 - 2s - loss: 2.6379 - accuracy: 0.1038 - val_loss: 2.7774 - val_accuracy: 0.0667 - 2s/epoch - 6ms/step
Epoch 5/10
283/283 - 2s - loss: 2.6381 - accuracy: 0.1042 - val_loss: 2.7913 - val_accuracy: 0.0667 - 2s/epoch - 6ms/step
Epoch 6/10
283/283 - 2s - loss: 2.6378 - accuracy: 0.1053 - val_loss: 2.7847 - val_accuracy: 0.0667 - 2s/epoch - 6ms/step
Epoch 7/10
283/283 - 2s - loss: 2.6373 - accuracy: 0.1025 - val_loss: 2.7908 - val_accuracy: 0.0667 - 2s/epoch - 6ms/step
Epoch 8/10
283/283 - 2s - loss: 2.6372 - accuracy: 0.1033 - val_loss: 2.8041 - val_accuracy: 0.0667 - 2s/epoch - 6ms/step
Epoch 9/10
283/283 - 2s - loss: 2.6376 - accuracy: 0.1041 - val_loss: 2.7971 - val_accuracy: 0.0667 - 2s/epoch - 6ms/step
Epoch 10/10
283/283 - 2s - loss: 2.6371 - accuracy: 0.1058 - val_loss: 2.7860 - val_accuracy: 0.0667 - 2s/epoch - 6ms/step
Baseline Error: 93.33%

df31.sort_values(by=['Validation Accuracy'],ascending=False)

	Model Name	Train Accuracy	Validation Accuracy	Train Loss	Validation Loss	History
1	LeNet	0.728622	0.576000	0.878840	1.332178	<keras.callbacks.History object at 0x7f5a0c5dd...
0	Base 31 Model	0.780128	0.537667	0.679253	1.505984	<keras.callbacks.History object at 0x7f5a1785d...
2	AlexNet	0.105782	0.066667	2.637130	2.785964	<keras.callbacks.History object at 0x7f5a0c3bb...

Model Name	Train Accuracy	Validation Accuracy	Train Loss	Validation Loss	Evaluation
Base 31 Model	78.11%	68.77%	0.6784	1.0055	Reasonable performance, slight overfitting.
LeNet	73.01%	57.37%	0.8786	1.3403	Signs of overfitting, might perform better with regularization.
AlexNet	10.58%	6.67%	2.6364	2.7977	Poor performance.

model = Lenet(128)
model.compile(optimizer='adam',
              loss= 'categorical_crossentropy',
              metrics=['accuracy'])
history = model.fit(X_train128, train_labels, validation_data=(X_val128, validation_labels),
                    epochs=10, batch_size=32, verbose=2, class_weight = weights)


scores = model.evaluate(X_val128, validation_labels, verbose=2)
print("Baseline Error: %.2f%%" % (100-scores[1]*100))
plot_history(history)
df128.loc[len(df128)] = ['LeNet', history.history['accuracy'][-1], history.history['val_accuracy'][-1], history.history['loss'][-1], history.history['val_loss'][-1], history]

Epoch 1/10
283/283 - 4s - loss: 1.7415 - accuracy: 0.4465 - val_loss: 1.4099 - val_accuracy: 0.5540 - 4s/epoch - 13ms/step
Epoch 2/10
283/283 - 1s - loss: 0.9844 - accuracy: 0.6986 - val_loss: 1.0954 - val_accuracy: 0.6657 - 1s/epoch - 5ms/step
Epoch 3/10
283/283 - 1s - loss: 0.5125 - accuracy: 0.8605 - val_loss: 1.0119 - val_accuracy: 0.6897 - 1s/epoch - 5ms/step
Epoch 4/10
283/283 - 1s - loss: 0.2301 - accuracy: 0.9485 - val_loss: 1.1692 - val_accuracy: 0.6587 - 1s/epoch - 5ms/step
Epoch 5/10
283/283 - 1s - loss: 0.0941 - accuracy: 0.9842 - val_loss: 1.0573 - val_accuracy: 0.7023 - 1s/epoch - 5ms/step
Epoch 6/10
283/283 - 1s - loss: 0.0371 - accuracy: 0.9963 - val_loss: 1.1030 - val_accuracy: 0.7083 - 1s/epoch - 5ms/step
Epoch 7/10
283/283 - 1s - loss: 0.0154 - accuracy: 0.9993 - val_loss: 1.1506 - val_accuracy: 0.7007 - 1s/epoch - 5ms/step
Epoch 8/10
283/283 - 1s - loss: 0.0083 - accuracy: 1.0000 - val_loss: 1.1598 - val_accuracy: 0.7057 - 1s/epoch - 5ms/step
Epoch 9/10
283/283 - 1s - loss: 0.0052 - accuracy: 1.0000 - val_loss: 1.1819 - val_accuracy: 0.7057 - 1s/epoch - 5ms/step
Epoch 10/10
283/283 - 1s - loss: 0.0036 - accuracy: 1.0000 - val_loss: 1.2037 - val_accuracy: 0.7040 - 1s/epoch - 5ms/step
94/94 - 0s - loss: 1.2037 - accuracy: 0.7040 - 265ms/epoch - 3ms/step
Baseline Error: 29.60%

model = AlexNet(128)
model.compile(optimizer='adam',
              loss= 'categorical_crossentropy',
              metrics=['accuracy'])
history = model.fit(X_train128, train_labels, validation_data=(X_val128, validation_labels),
                    epochs=10, batch_size=32, verbose=2, class_weight = weights)


scores = model.evaluate(X_val128, validation_labels, verbose=2)
print("Baseline Error: %.2f%%" % (100-scores[1]*100))
plot_history(history)
df128.loc[len(df128)] = ['AlexNet', history.history['accuracy'][-1], history.history['val_accuracy'][-1], history.history['loss'][-1], history.history['val_loss'][-1], history]

Epoch 1/10
283/283 - 8s - loss: 2.6561 - accuracy: 0.0963 - val_loss: 2.7892 - val_accuracy: 0.0667 - 8s/epoch - 28ms/step
Epoch 2/10
283/283 - 4s - loss: 2.6407 - accuracy: 0.1023 - val_loss: 2.7942 - val_accuracy: 0.0667 - 4s/epoch - 14ms/step
Epoch 3/10
283/283 - 4s - loss: 2.6387 - accuracy: 0.1002 - val_loss: 2.7831 - val_accuracy: 0.0667 - 4s/epoch - 14ms/step
Epoch 4/10
283/283 - 4s - loss: 2.6380 - accuracy: 0.1041 - val_loss: 2.7782 - val_accuracy: 0.0667 - 4s/epoch - 14ms/step
Epoch 5/10
283/283 - 4s - loss: 2.6379 - accuracy: 0.1042 - val_loss: 2.7914 - val_accuracy: 0.0667 - 4s/epoch - 14ms/step
Epoch 6/10
283/283 - 4s - loss: 2.6378 - accuracy: 0.1054 - val_loss: 2.7871 - val_accuracy: 0.0667 - 4s/epoch - 14ms/step
Epoch 7/10
283/283 - 4s - loss: 2.6373 - accuracy: 0.1025 - val_loss: 2.7928 - val_accuracy: 0.0667 - 4s/epoch - 14ms/step
Epoch 8/10
283/283 - 4s - loss: 2.6372 - accuracy: 0.1038 - val_loss: 2.8032 - val_accuracy: 0.0667 - 4s/epoch - 14ms/step
Epoch 9/10
283/283 - 4s - loss: 2.6376 - accuracy: 0.1040 - val_loss: 2.7959 - val_accuracy: 0.0667 - 4s/epoch - 14ms/step
Epoch 10/10
283/283 - 4s - loss: 2.6371 - accuracy: 0.1058 - val_loss: 2.7880 - val_accuracy: 0.0667 - 4s/epoch - 14ms/step
94/94 - 0s - loss: 2.7880 - accuracy: 0.0667 - 451ms/epoch - 5ms/step
Baseline Error: 93.33%

df128

	Model	Train Accuracy	Validation Accuracy	Train Loss	Validation Loss	History
0	Base 128 Model	0.985490	0.769333	0.045317	1.295857	<keras.callbacks.History object at 0x7f5a16d74...
1	LeNet	1.000000	0.704000	0.003636	1.203718	<keras.callbacks.History object at 0x7f59ec532...
2	AlexNet	0.105782	0.066667	2.637087	2.787983	<keras.callbacks.History object at 0x7f59ec79e...

Model	Train Accuracy	Validation Accuracy	Train Loss	Validation Loss	Evaluation
Base 128 Model	98.28%	76.63%	0.05399	1.32059	High accuracy, potential overfitting, check for regularization.
LeNet	98.90%	69.73%	0.10311	1.08369	High accuracy, potential overfitting regularization might be good.
AlexNet	10.58%	6.67%	2.63660	2.79447	Poor performance

Model Associated Improvements

Recall the previous markdown involving methods of improving our models, we can now try to implement some of these

1. Data Augmentation:

   • Description: Data augmentation involves applying random transformations to the training data, such as rotation, scaling, and flipping. This increases the diversity of the training set, helping the model generalize better to different variations of the input data.
   • Benefits: Mitigates overfitting by exposing the model to a wider range of variations in the data.

2. Batch Normalization:

   • Description: Batch normalization normalizes the input of each layer in a neural network to have zero mean and unit variance. This can accelerate training, improve model stability, and reduce sensitivity to weight initialization.
   • Benefits: Enhances convergence speed and can make the model less dependent on the choice of hyperparameters.

3. Learning Rate Scheduling:

   • Description: Adjusting the learning rate during training can be beneficial. For example, starting with a higher learning rate and gradually reducing it as training progresses can help the model converge more effectively.
   • Benefits: Can improve convergence, prevent overshooting, and result in a more stable training process.

4. Regularization Techniques:

   • Description: Techniques like dropout, L1, and L2 regularization can be applied to prevent overfitting. Dropout randomly drops units during training, while L1 and L2 regularization penalize large weights.
   • Benefits: Helps the model generalize better to new data and reduces the risk of overfitting.

5. Early Stopping:

   • Description: Monitor the validation loss during training and stop training when the loss stops improving. This prevents overfitting and saves computational resources.
   • Benefits: Guards against overfitting and ensures the model is not trained for too many epochs.

6. Adjusting Weights:

   • Description: We can adjust the weights in order to account for the imbalanced distribution of the dataset.
   • Benefits: Helps address the issue of imbalanced classes by assigning higher weights to under-represented classes during training. This ensures that the model pays more attention to minority classes, improving overall predictive performance and reducing bias towards the majority class.

Adjusting Weights

We can easily calculate the weights for each class by dividing the total number of images by the number of images in each class. We can then use these weights to adjust the loss function during training.

```python

#calculate class weights for each label

train_class['weights'] = (1/train_class['count']) * (sum(train_class['count'])/len(train_data.class_names))
class_weight = train_class['weights'].to_dict()

Regularization Techniques for model 31

As we are not actually using the LeNet and AlexNet models, we will not be implementing the regularization techniques for them. However we will be implementing them for the base model in order to try and improve the performance of the model. As a bonus, we should also use the class weights that we previously computed in order to accomodate for the imbalanced dataset.

# model_31 = models.Sequential([
#     layers.Conv2D(32, (3, 3), activation='relu', input_shape=(31, 31, 1)),
#     layers.MaxPooling2D((2, 2)),
#     layers.Conv2D(64, (3, 3), activation='relu'),
#     layers.MaxPooling2D((2, 2)),
#     layers.Conv2D(64, (3, 3), activation='relu'),
#     layers.Flatten(),
#     layers.Dense(64, activation='relu'),
#     layers.Dense(15, activation='softmax')
# ])

#this is the original "base model" that we used to overcome the overfitting we can apply regularization
# to improve teh accuracy we should also increase the number of epochs

model_31new = models.Sequential([
    layers.Conv2D(32, (3, 3), activation='relu', input_shape=(31, 31, 1)),
    layers.MaxPooling2D((2, 2)),
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.MaxPooling2D((2, 2)),
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.Flatten(),
    layers.Dense(64, activation='relu', kernel_regularizer=regularizers.l2(0.001)),
    layers.Dense(15, activation='softmax')
])

model_31new.compile(optimizer='adam',
                loss= 'categorical_crossentropy',
                metrics=['accuracy'])

history = model_31new.fit(X_train31, train_labels, validation_data=(X_val31, validation_labels),
                    epochs=100, batch_size=32, verbose=2, class_weight = class_weight)

scores = model_31new.evaluate(X_val31, validation_labels, verbose=2)
print("Baseline Error: %.2f%%" % (100-scores[1]*100))
plot_history(history)
df31.loc[len(df31)] = ['RegularizedModel', history.history['accuracy'][-1], history.history['val_accuracy'][-1], history.history['loss'][-1], history.history['val_loss'][-1], history]

Epoch 1/100
283/283 - 3s - loss: 2.9040 - accuracy: 0.1603 - val_loss: 2.6126 - val_accuracy: 0.1637 - 3s/epoch - 11ms/step
Epoch 2/100
283/283 - 1s - loss: 2.3496 - accuracy: 0.3056 - val_loss: 2.0725 - val_accuracy: 0.3460 - 1s/epoch - 4ms/step
Epoch 3/100
283/283 - 1s - loss: 1.9670 - accuracy: 0.4129 - val_loss: 1.8954 - val_accuracy: 0.3897 - 1s/epoch - 4ms/step
Epoch 4/100
283/283 - 1s - loss: 1.6449 - accuracy: 0.5177 - val_loss: 1.4937 - val_accuracy: 0.5353 - 1s/epoch - 4ms/step
Epoch 5/100
283/283 - 1s - loss: 1.4355 - accuracy: 0.5853 - val_loss: 1.6425 - val_accuracy: 0.5177 - 1s/epoch - 4ms/step
Epoch 6/100
283/283 - 1s - loss: 1.2782 - accuracy: 0.6371 - val_loss: 1.2124 - val_accuracy: 0.6377 - 1s/epoch - 4ms/step
Epoch 7/100
283/283 - 1s - loss: 1.1504 - accuracy: 0.6698 - val_loss: 1.1914 - val_accuracy: 0.6370 - 1s/epoch - 4ms/step
Epoch 8/100
283/283 - 1s - loss: 1.0742 - accuracy: 0.6893 - val_loss: 1.4247 - val_accuracy: 0.5690 - 1s/epoch - 4ms/step
Epoch 9/100
283/283 - 1s - loss: 1.0145 - accuracy: 0.7117 - val_loss: 1.1554 - val_accuracy: 0.6557 - 1s/epoch - 4ms/step
Epoch 10/100
283/283 - 1s - loss: 0.8961 - accuracy: 0.7467 - val_loss: 1.0680 - val_accuracy: 0.6840 - 1s/epoch - 4ms/step
Epoch 11/100
283/283 - 1s - loss: 0.8263 - accuracy: 0.7687 - val_loss: 1.0278 - val_accuracy: 0.7077 - 1s/epoch - 4ms/step
Epoch 12/100
283/283 - 1s - loss: 0.7837 - accuracy: 0.7794 - val_loss: 1.0226 - val_accuracy: 0.7207 - 1s/epoch - 4ms/step
Epoch 13/100
283/283 - 1s - loss: 0.7141 - accuracy: 0.8056 - val_loss: 1.0005 - val_accuracy: 0.7190 - 1s/epoch - 4ms/step
Epoch 14/100
283/283 - 1s - loss: 0.6749 - accuracy: 0.8136 - val_loss: 0.9839 - val_accuracy: 0.7360 - 1s/epoch - 4ms/step
Epoch 15/100
283/283 - 1s - loss: 0.6323 - accuracy: 0.8241 - val_loss: 0.9788 - val_accuracy: 0.7250 - 1s/epoch - 4ms/step
Epoch 16/100
283/283 - 1s - loss: 0.5862 - accuracy: 0.8419 - val_loss: 1.2402 - val_accuracy: 0.6657 - 1s/epoch - 4ms/step
Epoch 17/100
283/283 - 1s - loss: 0.5957 - accuracy: 0.8403 - val_loss: 0.8912 - val_accuracy: 0.7617 - 1s/epoch - 4ms/step
Epoch 18/100
283/283 - 1s - loss: 0.5359 - accuracy: 0.8604 - val_loss: 0.8845 - val_accuracy: 0.7753 - 1s/epoch - 4ms/step
Epoch 19/100
283/283 - 1s - loss: 0.4801 - accuracy: 0.8765 - val_loss: 1.0016 - val_accuracy: 0.7450 - 1s/epoch - 4ms/step
Epoch 20/100
283/283 - 1s - loss: 0.4763 - accuracy: 0.8759 - val_loss: 0.9191 - val_accuracy: 0.7763 - 1s/epoch - 5ms/step
Epoch 21/100
283/283 - 1s - loss: 0.4283 - accuracy: 0.8929 - val_loss: 0.9007 - val_accuracy: 0.7840 - 1s/epoch - 4ms/step
Epoch 22/100
283/283 - 1s - loss: 0.4200 - accuracy: 0.8965 - val_loss: 1.0420 - val_accuracy: 0.7533 - 1s/epoch - 4ms/step
Epoch 23/100
283/283 - 1s - loss: 0.4031 - accuracy: 0.8999 - val_loss: 1.0916 - val_accuracy: 0.7257 - 1s/epoch - 4ms/step
Epoch 24/100
283/283 - 1s - loss: 0.3992 - accuracy: 0.9010 - val_loss: 0.8774 - val_accuracy: 0.7943 - 1s/epoch - 4ms/step
Epoch 25/100
283/283 - 1s - loss: 0.3751 - accuracy: 0.9082 - val_loss: 0.9148 - val_accuracy: 0.7803 - 1s/epoch - 4ms/step
Epoch 26/100
283/283 - 1s - loss: 0.3647 - accuracy: 0.9170 - val_loss: 0.9116 - val_accuracy: 0.7920 - 1s/epoch - 4ms/step
Epoch 27/100
283/283 - 1s - loss: 0.3553 - accuracy: 0.9170 - val_loss: 0.9156 - val_accuracy: 0.7817 - 1s/epoch - 4ms/step
Epoch 28/100
283/283 - 1s - loss: 0.3553 - accuracy: 0.9222 - val_loss: 0.8456 - val_accuracy: 0.8017 - 1s/epoch - 4ms/step
Epoch 29/100
283/283 - 1s - loss: 0.2790 - accuracy: 0.9436 - val_loss: 0.9477 - val_accuracy: 0.7830 - 1s/epoch - 4ms/step
Epoch 30/100
283/283 - 1s - loss: 0.2785 - accuracy: 0.9417 - val_loss: 0.9205 - val_accuracy: 0.8020 - 1s/epoch - 4ms/step
Epoch 31/100
283/283 - 1s - loss: 0.2923 - accuracy: 0.9399 - val_loss: 0.9582 - val_accuracy: 0.7960 - 1s/epoch - 4ms/step
Epoch 32/100
283/283 - 1s - loss: 0.3059 - accuracy: 0.9363 - val_loss: 1.1586 - val_accuracy: 0.7507 - 1s/epoch - 4ms/step
Epoch 33/100
283/283 - 1s - loss: 0.2803 - accuracy: 0.9409 - val_loss: 1.0094 - val_accuracy: 0.7833 - 1s/epoch - 4ms/step
Epoch 34/100
283/283 - 1s - loss: 0.2728 - accuracy: 0.9428 - val_loss: 1.0678 - val_accuracy: 0.7720 - 1s/epoch - 4ms/step
Epoch 35/100
283/283 - 1s - loss: 0.2742 - accuracy: 0.9454 - val_loss: 0.9388 - val_accuracy: 0.8123 - 1s/epoch - 4ms/step
Epoch 36/100
283/283 - 1s - loss: 0.2674 - accuracy: 0.9476 - val_loss: 0.9678 - val_accuracy: 0.8003 - 1s/epoch - 4ms/step
Epoch 37/100
283/283 - 1s - loss: 0.2332 - accuracy: 0.9599 - val_loss: 1.0466 - val_accuracy: 0.7847 - 1s/epoch - 4ms/step
Epoch 38/100
283/283 - 1s - loss: 0.2561 - accuracy: 0.9497 - val_loss: 1.0662 - val_accuracy: 0.7863 - 1s/epoch - 4ms/step
Epoch 39/100
283/283 - 1s - loss: 0.2265 - accuracy: 0.9605 - val_loss: 1.0283 - val_accuracy: 0.8040 - 1s/epoch - 4ms/step
Epoch 40/100
283/283 - 1s - loss: 0.2234 - accuracy: 0.9610 - val_loss: 0.9668 - val_accuracy: 0.8257 - 1s/epoch - 4ms/step
Epoch 41/100
283/283 - 1s - loss: 0.2363 - accuracy: 0.9565 - val_loss: 1.3403 - val_accuracy: 0.7463 - 1s/epoch - 4ms/step
Epoch 42/100
283/283 - 1s - loss: 0.2284 - accuracy: 0.9593 - val_loss: 0.9845 - val_accuracy: 0.8140 - 1s/epoch - 4ms/step
Epoch 43/100
283/283 - 1s - loss: 0.2112 - accuracy: 0.9661 - val_loss: 1.0936 - val_accuracy: 0.8020 - 1s/epoch - 4ms/step
Epoch 44/100
283/283 - 1s - loss: 0.1736 - accuracy: 0.9782 - val_loss: 1.0207 - val_accuracy: 0.8233 - 1s/epoch - 4ms/step
Epoch 45/100
283/283 - 1s - loss: 0.2062 - accuracy: 0.9668 - val_loss: 1.1472 - val_accuracy: 0.7943 - 1s/epoch - 4ms/step
Epoch 46/100
283/283 - 1s - loss: 0.3048 - accuracy: 0.9364 - val_loss: 1.3788 - val_accuracy: 0.7447 - 1s/epoch - 4ms/step
Epoch 47/100
283/283 - 1s - loss: 0.2304 - accuracy: 0.9597 - val_loss: 1.1196 - val_accuracy: 0.7870 - 1s/epoch - 4ms/step
Epoch 48/100
283/283 - 1s - loss: 0.1547 - accuracy: 0.9863 - val_loss: 0.9935 - val_accuracy: 0.8313 - 1s/epoch - 4ms/step
Epoch 49/100
283/283 - 1s - loss: 0.1456 - accuracy: 0.9888 - val_loss: 0.9410 - val_accuracy: 0.8340 - 1s/epoch - 4ms/step
Epoch 50/100
283/283 - 1s - loss: 0.2674 - accuracy: 0.9458 - val_loss: 1.0468 - val_accuracy: 0.8030 - 1s/epoch - 4ms/step
Epoch 51/100
283/283 - 1s - loss: 0.2777 - accuracy: 0.9474 - val_loss: 0.9536 - val_accuracy: 0.8293 - 1s/epoch - 4ms/step
Epoch 52/100
283/283 - 1s - loss: 0.1534 - accuracy: 0.9866 - val_loss: 0.9191 - val_accuracy: 0.8373 - 1s/epoch - 4ms/step
Epoch 53/100
283/283 - 1s - loss: 0.1385 - accuracy: 0.9908 - val_loss: 0.9717 - val_accuracy: 0.8390 - 1s/epoch - 4ms/step
Epoch 54/100
283/283 - 1s - loss: 0.2053 - accuracy: 0.9700 - val_loss: 1.3681 - val_accuracy: 0.7437 - 1s/epoch - 4ms/step
Epoch 55/100
283/283 - 1s - loss: 0.2513 - accuracy: 0.9529 - val_loss: 1.0173 - val_accuracy: 0.8163 - 1s/epoch - 4ms/step
Epoch 56/100
283/283 - 1s - loss: 0.1437 - accuracy: 0.9874 - val_loss: 1.0223 - val_accuracy: 0.8287 - 1s/epoch - 4ms/step
Epoch 57/100
283/283 - 1s - loss: 0.1913 - accuracy: 0.9708 - val_loss: 1.1192 - val_accuracy: 0.8100 - 1s/epoch - 4ms/step
Epoch 58/100
283/283 - 1s - loss: 0.2272 - accuracy: 0.9584 - val_loss: 1.0534 - val_accuracy: 0.8147 - 1s/epoch - 4ms/step
Epoch 59/100
283/283 - 1s - loss: 0.2448 - accuracy: 0.9540 - val_loss: 1.0917 - val_accuracy: 0.8117 - 1s/epoch - 4ms/step
Epoch 60/100
283/283 - 1s - loss: 0.1365 - accuracy: 0.9908 - val_loss: 1.0214 - val_accuracy: 0.8377 - 1s/epoch - 4ms/step
Epoch 61/100
283/283 - 1s - loss: 0.1108 - accuracy: 0.9990 - val_loss: 0.9741 - val_accuracy: 0.8513 - 1s/epoch - 4ms/step
Epoch 62/100
283/283 - 1s - loss: 0.1058 - accuracy: 0.9982 - val_loss: 0.9876 - val_accuracy: 0.8457 - 1s/epoch - 4ms/step
Epoch 63/100
283/283 - 1s - loss: 0.2986 - accuracy: 0.9371 - val_loss: 1.1669 - val_accuracy: 0.7910 - 1s/epoch - 4ms/step
Epoch 64/100
283/283 - 1s - loss: 0.1726 - accuracy: 0.9764 - val_loss: 1.0821 - val_accuracy: 0.8123 - 1s/epoch - 4ms/step
Epoch 65/100
283/283 - 1s - loss: 0.2194 - accuracy: 0.9616 - val_loss: 0.9538 - val_accuracy: 0.8303 - 1s/epoch - 4ms/step
Epoch 66/100
283/283 - 1s - loss: 0.1297 - accuracy: 0.9897 - val_loss: 1.0374 - val_accuracy: 0.8263 - 1s/epoch - 4ms/step
Epoch 67/100
283/283 - 1s - loss: 0.1989 - accuracy: 0.9668 - val_loss: 1.2006 - val_accuracy: 0.7837 - 1s/epoch - 4ms/step
Epoch 68/100
283/283 - 1s - loss: 0.1846 - accuracy: 0.9714 - val_loss: 1.1019 - val_accuracy: 0.8040 - 1s/epoch - 4ms/step
Epoch 69/100
283/283 - 1s - loss: 0.1686 - accuracy: 0.9768 - val_loss: 1.0287 - val_accuracy: 0.8280 - 1s/epoch - 4ms/step
Epoch 70/100
283/283 - 1s - loss: 0.1376 - accuracy: 0.9876 - val_loss: 1.4380 - val_accuracy: 0.7593 - 1s/epoch - 4ms/step
Epoch 71/100
283/283 - 1s - loss: 0.3305 - accuracy: 0.9399 - val_loss: 1.0856 - val_accuracy: 0.8107 - 1s/epoch - 4ms/step
Epoch 72/100
283/283 - 1s - loss: 0.1367 - accuracy: 0.9879 - val_loss: 1.0187 - val_accuracy: 0.8247 - 1s/epoch - 4ms/step
Epoch 73/100
283/283 - 1s - loss: 0.1077 - accuracy: 0.9977 - val_loss: 0.9249 - val_accuracy: 0.8497 - 1s/epoch - 4ms/step
Epoch 74/100
283/283 - 1s - loss: 0.0960 - accuracy: 0.9997 - val_loss: 0.9571 - val_accuracy: 0.8480 - 1s/epoch - 4ms/step
Epoch 75/100
283/283 - 1s - loss: 0.0896 - accuracy: 0.9999 - val_loss: 0.9460 - val_accuracy: 0.8503 - 1s/epoch - 4ms/step
Epoch 76/100
283/283 - 1s - loss: 0.0839 - accuracy: 1.0000 - val_loss: 0.9172 - val_accuracy: 0.8530 - 1s/epoch - 4ms/step
Epoch 77/100
283/283 - 1s - loss: 0.1749 - accuracy: 0.9756 - val_loss: 1.4209 - val_accuracy: 0.7100 - 1s/epoch - 4ms/step
Epoch 78/100
283/283 - 1s - loss: 0.3389 - accuracy: 0.9157 - val_loss: 0.9443 - val_accuracy: 0.8233 - 1s/epoch - 4ms/step
Epoch 79/100
283/283 - 1s - loss: 0.1689 - accuracy: 0.9719 - val_loss: 1.0381 - val_accuracy: 0.8113 - 1s/epoch - 4ms/step
Epoch 80/100
283/283 - 1s - loss: 0.1420 - accuracy: 0.9838 - val_loss: 1.0395 - val_accuracy: 0.8057 - 1s/epoch - 4ms/step
Epoch 81/100
283/283 - 1s - loss: 0.1464 - accuracy: 0.9807 - val_loss: 0.9728 - val_accuracy: 0.8333 - 1s/epoch - 4ms/step
Epoch 82/100
283/283 - 1s - loss: 0.2286 - accuracy: 0.9566 - val_loss: 1.0039 - val_accuracy: 0.8113 - 1s/epoch - 4ms/step
Epoch 83/100
283/283 - 1s - loss: 0.1534 - accuracy: 0.9793 - val_loss: 0.9491 - val_accuracy: 0.8370 - 1s/epoch - 4ms/step
Epoch 84/100
283/283 - 1s - loss: 0.0978 - accuracy: 0.9986 - val_loss: 0.9322 - val_accuracy: 0.8480 - 1s/epoch - 4ms/step
Epoch 85/100
283/283 - 1s - loss: 0.0874 - accuracy: 0.9999 - val_loss: 0.9261 - val_accuracy: 0.8523 - 1s/epoch - 4ms/step
Epoch 86/100
283/283 - 1s - loss: 0.0815 - accuracy: 0.9999 - val_loss: 0.9294 - val_accuracy: 0.8523 - 1s/epoch - 4ms/step
Epoch 87/100
283/283 - 1s - loss: 0.0762 - accuracy: 1.0000 - val_loss: 0.9334 - val_accuracy: 0.8480 - 1s/epoch - 4ms/step
Epoch 88/100
283/283 - 1s - loss: 0.0710 - accuracy: 1.0000 - val_loss: 0.9225 - val_accuracy: 0.8507 - 1s/epoch - 4ms/step
Epoch 89/100
283/283 - 1s - loss: 0.0671 - accuracy: 0.9996 - val_loss: 0.9600 - val_accuracy: 0.8433 - 1s/epoch - 4ms/step
Epoch 90/100
283/283 - 1s - loss: 0.4290 - accuracy: 0.8913 - val_loss: 0.9699 - val_accuracy: 0.8000 - 1s/epoch - 4ms/step
Epoch 91/100
283/283 - 1s - loss: 0.1875 - accuracy: 0.9621 - val_loss: 1.0075 - val_accuracy: 0.8177 - 1s/epoch - 4ms/step
Epoch 92/100
283/283 - 1s - loss: 0.1202 - accuracy: 0.9874 - val_loss: 0.8819 - val_accuracy: 0.8357 - 1s/epoch - 4ms/step
Epoch 93/100
283/283 - 1s - loss: 0.0878 - accuracy: 0.9986 - val_loss: 0.9248 - val_accuracy: 0.8450 - 1s/epoch - 4ms/step
Epoch 94/100
283/283 - 1s - loss: 0.0786 - accuracy: 0.9994 - val_loss: 0.8921 - val_accuracy: 0.8533 - 1s/epoch - 4ms/step
Epoch 95/100
283/283 - 1s - loss: 0.0720 - accuracy: 0.9997 - val_loss: 0.8870 - val_accuracy: 0.8563 - 1s/epoch - 4ms/step
Epoch 96/100
283/283 - 1s - loss: 0.0663 - accuracy: 1.0000 - val_loss: 0.8915 - val_accuracy: 0.8570 - 1s/epoch - 4ms/step
Epoch 97/100
283/283 - 1s - loss: 0.0612 - accuracy: 1.0000 - val_loss: 0.8999 - val_accuracy: 0.8523 - 1s/epoch - 4ms/step
Epoch 98/100
283/283 - 1s - loss: 0.0569 - accuracy: 1.0000 - val_loss: 0.8977 - val_accuracy: 0.8527 - 1s/epoch - 4ms/step
Epoch 99/100
283/283 - 1s - loss: 0.4509 - accuracy: 0.8892 - val_loss: 1.1631 - val_accuracy: 0.7547 - 1s/epoch - 4ms/step
Epoch 100/100
283/283 - 1s - loss: 0.1816 - accuracy: 0.9638 - val_loss: 1.0630 - val_accuracy: 0.8053 - 1s/epoch - 4ms/step
94/94 - 0s - loss: 1.0630 - accuracy: 0.8053 - 196ms/epoch - 2ms/step
Baseline Error: 19.47%

As seen from the graph, the learning of the model starts decreasing after around 30 epochs, we can try a learning rate scheduler to see if we can improve the performance of the model. As there is still some overfitting, we can also try to implement dropout in order to reduce the overfitting.

As the 31x31 images are simple, we will try not to overcomplicate the model for now and look for easier solutions to this problem.

#First test with dropout


model_31new = models.Sequential([
    layers.Conv2D(32, (3, 3), activation='relu', input_shape=(31, 31, 1)),
    layers.MaxPooling2D((2, 2)),
    
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.MaxPooling2D((2, 2)),
    
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.MaxPooling2D((2, 2)),
    
    layers.Flatten(),
    
    layers.Dense(64, activation='relu', kernel_regularizer=regularizers.l2(0.001)),
    layers.Dropout(0.5),  # Add dropout with dropout rate of 0.5
    
    layers.Dense(15, activation='softmax')
])


model_31new.compile(optimizer='adam',
                loss= 'categorical_crossentropy',
                metrics=['accuracy'])

history = model_31new.fit(X_train31, train_labels, validation_data=(X_val31, validation_labels),
                    epochs=100, batch_size=32, verbose=2, class_weight = class_weight)

scores = model_31new.evaluate(X_val31, validation_labels, verbose=2)
print("Baseline Error: %.2f%%" % (100-scores[1]*100))
plot_history(history)
df31.loc[len(df31)] = ['RegularizedDOModel', history.history['accuracy'][-1], history.history['val_accuracy'][-1], history.history['loss'][-1], history.history['val_loss'][-1], history]

Epoch 1/100
283/283 - 3s - loss: 3.0926 - accuracy: 0.1304 - val_loss: 2.6872 - val_accuracy: 0.1233 - 3s/epoch - 11ms/step
Epoch 2/100
283/283 - 1s - loss: 2.6139 - accuracy: 0.2318 - val_loss: 2.3090 - val_accuracy: 0.2317 - 1s/epoch - 4ms/step
Epoch 3/100
283/283 - 1s - loss: 2.3558 - accuracy: 0.2888 - val_loss: 2.2681 - val_accuracy: 0.2803 - 1s/epoch - 4ms/step
Epoch 4/100
283/283 - 1s - loss: 2.1560 - accuracy: 0.3432 - val_loss: 1.9635 - val_accuracy: 0.3587 - 1s/epoch - 4ms/step
Epoch 5/100
283/283 - 1s - loss: 2.0298 - accuracy: 0.3739 - val_loss: 1.8783 - val_accuracy: 0.3883 - 1s/epoch - 4ms/step
Epoch 6/100
283/283 - 1s - loss: 1.9356 - accuracy: 0.3953 - val_loss: 1.7106 - val_accuracy: 0.4420 - 1s/epoch - 4ms/step
Epoch 7/100
283/283 - 1s - loss: 1.8030 - accuracy: 0.4366 - val_loss: 1.7225 - val_accuracy: 0.4383 - 1s/epoch - 4ms/step
Epoch 8/100
283/283 - 1s - loss: 1.7490 - accuracy: 0.4527 - val_loss: 1.6495 - val_accuracy: 0.4647 - 1s/epoch - 4ms/step
Epoch 9/100
283/283 - 1s - loss: 1.6968 - accuracy: 0.4662 - val_loss: 1.6096 - val_accuracy: 0.4760 - 1s/epoch - 4ms/step
Epoch 10/100
283/283 - 1s - loss: 1.5971 - accuracy: 0.4955 - val_loss: 1.4843 - val_accuracy: 0.5170 - 1s/epoch - 4ms/step
Epoch 11/100
283/283 - 1s - loss: 1.5600 - accuracy: 0.5047 - val_loss: 1.4185 - val_accuracy: 0.5483 - 1s/epoch - 4ms/step
Epoch 12/100
283/283 - 1s - loss: 1.5242 - accuracy: 0.5207 - val_loss: 1.4550 - val_accuracy: 0.5307 - 1s/epoch - 4ms/step
Epoch 13/100
283/283 - 1s - loss: 1.4785 - accuracy: 0.5268 - val_loss: 1.3157 - val_accuracy: 0.5787 - 1s/epoch - 4ms/step
Epoch 14/100
283/283 - 1s - loss: 1.4198 - accuracy: 0.5457 - val_loss: 1.4691 - val_accuracy: 0.5463 - 1s/epoch - 4ms/step
Epoch 15/100
283/283 - 1s - loss: 1.3855 - accuracy: 0.5541 - val_loss: 1.2320 - val_accuracy: 0.6107 - 1s/epoch - 4ms/step
Epoch 16/100
283/283 - 1s - loss: 1.3885 - accuracy: 0.5575 - val_loss: 1.3340 - val_accuracy: 0.5907 - 1s/epoch - 4ms/step
Epoch 17/100
283/283 - 1s - loss: 1.3450 - accuracy: 0.5697 - val_loss: 1.1889 - val_accuracy: 0.6267 - 1s/epoch - 4ms/step
Epoch 18/100
283/283 - 1s - loss: 1.3016 - accuracy: 0.5881 - val_loss: 1.2507 - val_accuracy: 0.6163 - 1s/epoch - 4ms/step
Epoch 19/100
283/283 - 1s - loss: 1.2828 - accuracy: 0.5893 - val_loss: 1.2349 - val_accuracy: 0.6093 - 1s/epoch - 4ms/step
Epoch 20/100
283/283 - 1s - loss: 1.2424 - accuracy: 0.5951 - val_loss: 1.1769 - val_accuracy: 0.6363 - 1s/epoch - 4ms/step
Epoch 21/100
283/283 - 1s - loss: 1.2252 - accuracy: 0.6051 - val_loss: 1.2000 - val_accuracy: 0.6190 - 1s/epoch - 4ms/step
Epoch 22/100
283/283 - 1s - loss: 1.1951 - accuracy: 0.6180 - val_loss: 1.1418 - val_accuracy: 0.6477 - 1s/epoch - 4ms/step
Epoch 23/100
283/283 - 1s - loss: 1.1619 - accuracy: 0.6303 - val_loss: 1.2412 - val_accuracy: 0.6187 - 1s/epoch - 4ms/step
Epoch 24/100
283/283 - 1s - loss: 1.1772 - accuracy: 0.6273 - val_loss: 1.2253 - val_accuracy: 0.6217 - 1s/epoch - 4ms/step
Epoch 25/100
283/283 - 1s - loss: 1.1274 - accuracy: 0.6352 - val_loss: 1.0877 - val_accuracy: 0.6797 - 1s/epoch - 4ms/step
Epoch 26/100
283/283 - 1s - loss: 1.1090 - accuracy: 0.6379 - val_loss: 1.0402 - val_accuracy: 0.6817 - 1s/epoch - 4ms/step
Epoch 27/100
283/283 - 1s - loss: 1.0811 - accuracy: 0.6562 - val_loss: 1.1028 - val_accuracy: 0.6670 - 1s/epoch - 4ms/step
Epoch 28/100
283/283 - 1s - loss: 1.0673 - accuracy: 0.6564 - val_loss: 1.1353 - val_accuracy: 0.6457 - 1s/epoch - 4ms/step
Epoch 29/100
283/283 - 1s - loss: 1.0493 - accuracy: 0.6556 - val_loss: 1.0969 - val_accuracy: 0.6747 - 1s/epoch - 4ms/step
Epoch 30/100
283/283 - 1s - loss: 1.0140 - accuracy: 0.6688 - val_loss: 1.1082 - val_accuracy: 0.6630 - 1s/epoch - 5ms/step
Epoch 31/100
283/283 - 1s - loss: 0.9977 - accuracy: 0.6778 - val_loss: 1.0445 - val_accuracy: 0.6903 - 1s/epoch - 4ms/step
Epoch 32/100
283/283 - 1s - loss: 0.9672 - accuracy: 0.6871 - val_loss: 1.0117 - val_accuracy: 0.6937 - 1s/epoch - 4ms/step
Epoch 33/100
283/283 - 1s - loss: 0.9646 - accuracy: 0.6921 - val_loss: 1.0493 - val_accuracy: 0.6977 - 1s/epoch - 4ms/step
Epoch 34/100
283/283 - 1s - loss: 0.9615 - accuracy: 0.6951 - val_loss: 1.1128 - val_accuracy: 0.6733 - 1s/epoch - 4ms/step
Epoch 35/100
283/283 - 1s - loss: 0.9295 - accuracy: 0.7015 - val_loss: 1.0132 - val_accuracy: 0.7107 - 1s/epoch - 4ms/step
Epoch 36/100
283/283 - 1s - loss: 0.9292 - accuracy: 0.7010 - val_loss: 0.9217 - val_accuracy: 0.7257 - 1s/epoch - 4ms/step
Epoch 37/100
283/283 - 1s - loss: 0.9021 - accuracy: 0.7097 - val_loss: 1.0470 - val_accuracy: 0.7023 - 1s/epoch - 4ms/step
Epoch 38/100
283/283 - 1s - loss: 0.8687 - accuracy: 0.7198 - val_loss: 1.1369 - val_accuracy: 0.6720 - 1s/epoch - 4ms/step
Epoch 39/100
283/283 - 1s - loss: 0.8561 - accuracy: 0.7165 - val_loss: 0.9366 - val_accuracy: 0.7327 - 1s/epoch - 4ms/step
Epoch 40/100
283/283 - 1s - loss: 0.8414 - accuracy: 0.7255 - val_loss: 1.0177 - val_accuracy: 0.7077 - 1s/epoch - 4ms/step
Epoch 41/100
283/283 - 1s - loss: 0.8514 - accuracy: 0.7288 - val_loss: 1.0834 - val_accuracy: 0.7003 - 1s/epoch - 4ms/step
Epoch 42/100
283/283 - 1s - loss: 0.8504 - accuracy: 0.7234 - val_loss: 0.9322 - val_accuracy: 0.7330 - 1s/epoch - 4ms/step
Epoch 43/100
283/283 - 1s - loss: 0.8183 - accuracy: 0.7267 - val_loss: 0.9006 - val_accuracy: 0.7480 - 1s/epoch - 4ms/step
Epoch 44/100
283/283 - 1s - loss: 0.7856 - accuracy: 0.7473 - val_loss: 1.0261 - val_accuracy: 0.7167 - 1s/epoch - 4ms/step
Epoch 45/100
283/283 - 1s - loss: 0.8011 - accuracy: 0.7429 - val_loss: 0.9595 - val_accuracy: 0.7387 - 1s/epoch - 4ms/step
Epoch 46/100
283/283 - 1s - loss: 0.7876 - accuracy: 0.7430 - val_loss: 1.0160 - val_accuracy: 0.7213 - 1s/epoch - 4ms/step
Epoch 47/100
283/283 - 1s - loss: 0.7545 - accuracy: 0.7552 - val_loss: 0.9526 - val_accuracy: 0.7383 - 1s/epoch - 4ms/step
Epoch 48/100
283/283 - 1s - loss: 0.7515 - accuracy: 0.7607 - val_loss: 0.9214 - val_accuracy: 0.7490 - 1s/epoch - 4ms/step
Epoch 49/100
283/283 - 1s - loss: 0.7211 - accuracy: 0.7654 - val_loss: 0.8697 - val_accuracy: 0.7640 - 1s/epoch - 4ms/step
Epoch 50/100
283/283 - 1s - loss: 0.7012 - accuracy: 0.7719 - val_loss: 1.0443 - val_accuracy: 0.7317 - 1s/epoch - 4ms/step
Epoch 51/100
283/283 - 1s - loss: 0.7155 - accuracy: 0.7688 - val_loss: 0.8909 - val_accuracy: 0.7627 - 1s/epoch - 4ms/step
Epoch 52/100
283/283 - 1s - loss: 0.6987 - accuracy: 0.7741 - val_loss: 0.8303 - val_accuracy: 0.7783 - 1s/epoch - 4ms/step
Epoch 53/100
283/283 - 1s - loss: 0.6922 - accuracy: 0.7714 - val_loss: 0.9513 - val_accuracy: 0.7520 - 1s/epoch - 4ms/step
Epoch 54/100
283/283 - 1s - loss: 0.6978 - accuracy: 0.7737 - val_loss: 0.9404 - val_accuracy: 0.7550 - 1s/epoch - 4ms/step
Epoch 55/100
283/283 - 1s - loss: 0.6735 - accuracy: 0.7852 - val_loss: 1.1237 - val_accuracy: 0.7180 - 1s/epoch - 4ms/step
Epoch 56/100
283/283 - 1s - loss: 0.6302 - accuracy: 0.7966 - val_loss: 0.8594 - val_accuracy: 0.7753 - 1s/epoch - 4ms/step
Epoch 57/100
283/283 - 1s - loss: 0.6673 - accuracy: 0.7863 - val_loss: 0.8773 - val_accuracy: 0.7753 - 1s/epoch - 4ms/step
Epoch 58/100
283/283 - 1s - loss: 0.6471 - accuracy: 0.7890 - val_loss: 0.9261 - val_accuracy: 0.7530 - 1s/epoch - 4ms/step
Epoch 59/100
283/283 - 1s - loss: 0.6455 - accuracy: 0.7939 - val_loss: 1.0639 - val_accuracy: 0.7320 - 1s/epoch - 4ms/step
Epoch 60/100
283/283 - 1s - loss: 0.6480 - accuracy: 0.7905 - val_loss: 1.0014 - val_accuracy: 0.7507 - 1s/epoch - 4ms/step
Epoch 61/100
283/283 - 1s - loss: 0.6005 - accuracy: 0.8058 - val_loss: 0.9098 - val_accuracy: 0.7720 - 1s/epoch - 4ms/step
Epoch 62/100
283/283 - 1s - loss: 0.6057 - accuracy: 0.8063 - val_loss: 0.9143 - val_accuracy: 0.7720 - 1s/epoch - 4ms/step
Epoch 63/100
283/283 - 1s - loss: 0.6064 - accuracy: 0.8022 - val_loss: 0.8476 - val_accuracy: 0.7910 - 1s/epoch - 4ms/step
Epoch 64/100
283/283 - 1s - loss: 0.5839 - accuracy: 0.8138 - val_loss: 0.9319 - val_accuracy: 0.7783 - 1s/epoch - 4ms/step
Epoch 65/100
283/283 - 1s - loss: 0.5750 - accuracy: 0.8166 - val_loss: 1.0177 - val_accuracy: 0.7583 - 1s/epoch - 4ms/step
Epoch 66/100
283/283 - 1s - loss: 0.5767 - accuracy: 0.8207 - val_loss: 0.8622 - val_accuracy: 0.7900 - 1s/epoch - 4ms/step
Epoch 67/100
283/283 - 1s - loss: 0.5550 - accuracy: 0.8196 - val_loss: 0.9330 - val_accuracy: 0.7763 - 1s/epoch - 4ms/step
Epoch 68/100
283/283 - 1s - loss: 0.5714 - accuracy: 0.8192 - val_loss: 0.9594 - val_accuracy: 0.7690 - 1s/epoch - 4ms/step
Epoch 69/100
283/283 - 1s - loss: 0.5510 - accuracy: 0.8209 - val_loss: 0.8871 - val_accuracy: 0.7787 - 1s/epoch - 4ms/step
Epoch 70/100
283/283 - 1s - loss: 0.5407 - accuracy: 0.8251 - val_loss: 1.2404 - val_accuracy: 0.7157 - 1s/epoch - 4ms/step
Epoch 71/100
283/283 - 1s - loss: 0.6426 - accuracy: 0.8011 - val_loss: 1.0023 - val_accuracy: 0.7730 - 1s/epoch - 4ms/step
Epoch 72/100
283/283 - 1s - loss: 0.5180 - accuracy: 0.8343 - val_loss: 0.8859 - val_accuracy: 0.7833 - 1s/epoch - 4ms/step
Epoch 73/100
283/283 - 1s - loss: 0.5204 - accuracy: 0.8336 - val_loss: 0.9411 - val_accuracy: 0.7783 - 1s/epoch - 4ms/step
Epoch 74/100
283/283 - 1s - loss: 0.4945 - accuracy: 0.8405 - val_loss: 0.9288 - val_accuracy: 0.7840 - 1s/epoch - 4ms/step
Epoch 75/100
283/283 - 1s - loss: 0.5020 - accuracy: 0.8424 - val_loss: 0.9603 - val_accuracy: 0.7800 - 1s/epoch - 4ms/step
Epoch 76/100
283/283 - 1s - loss: 0.5003 - accuracy: 0.8437 - val_loss: 0.9257 - val_accuracy: 0.7910 - 1s/epoch - 4ms/step
Epoch 77/100
283/283 - 1s - loss: 0.5382 - accuracy: 0.8320 - val_loss: 0.9512 - val_accuracy: 0.7727 - 1s/epoch - 4ms/step
Epoch 78/100
283/283 - 1s - loss: 0.4734 - accuracy: 0.8468 - val_loss: 1.1122 - val_accuracy: 0.7623 - 1s/epoch - 4ms/step
Epoch 79/100
283/283 - 1s - loss: 0.5029 - accuracy: 0.8415 - val_loss: 0.9442 - val_accuracy: 0.7817 - 1s/epoch - 4ms/step
Epoch 80/100
283/283 - 1s - loss: 0.5069 - accuracy: 0.8411 - val_loss: 0.9137 - val_accuracy: 0.7953 - 1s/epoch - 4ms/step
Epoch 81/100
283/283 - 1s - loss: 0.4824 - accuracy: 0.8494 - val_loss: 0.8661 - val_accuracy: 0.7980 - 1s/epoch - 4ms/step
Epoch 82/100
283/283 - 1s - loss: 0.4738 - accuracy: 0.8504 - val_loss: 0.9566 - val_accuracy: 0.7937 - 1s/epoch - 4ms/step
Epoch 83/100
283/283 - 1s - loss: 0.4623 - accuracy: 0.8535 - val_loss: 1.2196 - val_accuracy: 0.7447 - 1s/epoch - 4ms/step
Epoch 84/100
283/283 - 1s - loss: 0.4770 - accuracy: 0.8491 - val_loss: 0.8735 - val_accuracy: 0.8027 - 1s/epoch - 4ms/step
Epoch 85/100
283/283 - 1s - loss: 0.4754 - accuracy: 0.8537 - val_loss: 0.9473 - val_accuracy: 0.7897 - 1s/epoch - 4ms/step
Epoch 86/100
283/283 - 1s - loss: 0.4673 - accuracy: 0.8576 - val_loss: 1.1136 - val_accuracy: 0.7593 - 1s/epoch - 4ms/step
Epoch 87/100
283/283 - 1s - loss: 0.4497 - accuracy: 0.8571 - val_loss: 0.9502 - val_accuracy: 0.7930 - 1s/epoch - 4ms/step
Epoch 88/100
283/283 - 1s - loss: 0.4612 - accuracy: 0.8535 - val_loss: 1.0115 - val_accuracy: 0.7653 - 1s/epoch - 4ms/step
Epoch 89/100
283/283 - 1s - loss: 0.4450 - accuracy: 0.8684 - val_loss: 1.0254 - val_accuracy: 0.7927 - 1s/epoch - 4ms/step
Epoch 90/100
283/283 - 1s - loss: 0.4521 - accuracy: 0.8609 - val_loss: 1.0812 - val_accuracy: 0.7770 - 1s/epoch - 4ms/step
Epoch 91/100
283/283 - 1s - loss: 0.4239 - accuracy: 0.8698 - val_loss: 0.9747 - val_accuracy: 0.8037 - 1s/epoch - 4ms/step
Epoch 92/100
283/283 - 1s - loss: 0.4473 - accuracy: 0.8626 - val_loss: 1.2167 - val_accuracy: 0.7530 - 1s/epoch - 4ms/step
Epoch 93/100
283/283 - 1s - loss: 0.4190 - accuracy: 0.8698 - val_loss: 1.0829 - val_accuracy: 0.7883 - 1s/epoch - 4ms/step
Epoch 94/100
283/283 - 1s - loss: 0.4217 - accuracy: 0.8694 - val_loss: 0.9980 - val_accuracy: 0.8003 - 1s/epoch - 4ms/step
Epoch 95/100
283/283 - 1s - loss: 0.4150 - accuracy: 0.8718 - val_loss: 1.0828 - val_accuracy: 0.7860 - 1s/epoch - 4ms/step
Epoch 96/100
283/283 - 1s - loss: 0.3943 - accuracy: 0.8833 - val_loss: 0.9102 - val_accuracy: 0.8160 - 1s/epoch - 5ms/step
Epoch 97/100
283/283 - 1s - loss: 0.4116 - accuracy: 0.8705 - val_loss: 0.9355 - val_accuracy: 0.7990 - 1s/epoch - 5ms/step
Epoch 98/100
283/283 - 1s - loss: 0.3878 - accuracy: 0.8784 - val_loss: 0.9674 - val_accuracy: 0.8057 - 1s/epoch - 4ms/step
Epoch 99/100
283/283 - 1s - loss: 0.4008 - accuracy: 0.8795 - val_loss: 1.2376 - val_accuracy: 0.7617 - 1s/epoch - 4ms/step
Epoch 100/100
283/283 - 1s - loss: 0.4130 - accuracy: 0.8704 - val_loss: 1.0561 - val_accuracy: 0.7900 - 1s/epoch - 4ms/step
94/94 - 0s - loss: 1.0561 - accuracy: 0.7900 - 218ms/epoch - 2ms/step
Baseline Error: 21.00%

We can see that the overfitting problem is basically non existent now, we can try to increase the epochs to let the model fully converge and see if we can get better results.

#Same model and other params, change epochs


model_31new = models.Sequential([
    layers.Conv2D(32, (3, 3), activation='relu', input_shape=(31, 31, 1)),
    layers.MaxPooling2D((2, 2)),
    
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.MaxPooling2D((2, 2)),
    
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.MaxPooling2D((2, 2)),
    
    layers.Flatten(),
    
    layers.Dense(64, activation='relu', kernel_regularizer=regularizers.l2(0.001)),
    layers.Dropout(0.5), 
    
    layers.Dense(15, activation='softmax')
])


model_31new.compile(optimizer='adam',
                loss= 'categorical_crossentropy',
                metrics=['accuracy'])

history = model_31new.fit(X_train31, train_labels, validation_data=(X_val31, validation_labels),
                    epochs=300, batch_size=32, verbose=2, class_weight = class_weight) #adjust epochs to 300

scores = model_31new.evaluate(X_val31, validation_labels, verbose=2)
print("Baseline Error: %.2f%%" % (100-scores[1]*100))
plot_history(history)
df31.loc[len(df31)] = ['RegularizedDOModel2', history.history['accuracy'][-1], history.history['val_accuracy'][-1], history.history['loss'][-1], history.history['val_loss'][-1], history]

Epoch 1/300
283/283 - 3s - loss: 3.1348 - accuracy: 0.1166 - val_loss: 2.8202 - val_accuracy: 0.0950 - 3s/epoch - 12ms/step
Epoch 2/300
283/283 - 1s - loss: 2.7163 - accuracy: 0.2100 - val_loss: 2.4495 - val_accuracy: 0.1860 - 1s/epoch - 4ms/step
Epoch 3/300
283/283 - 1s - loss: 2.4611 - accuracy: 0.2524 - val_loss: 2.3568 - val_accuracy: 0.2273 - 1s/epoch - 4ms/step
Epoch 4/300
283/283 - 1s - loss: 2.3119 - accuracy: 0.2952 - val_loss: 2.0924 - val_accuracy: 0.3103 - 1s/epoch - 4ms/step
Epoch 5/300
283/283 - 1s - loss: 2.1461 - accuracy: 0.3343 - val_loss: 1.9254 - val_accuracy: 0.3757 - 1s/epoch - 4ms/step
Epoch 6/300
283/283 - 1s - loss: 2.0484 - accuracy: 0.3665 - val_loss: 1.8364 - val_accuracy: 0.4033 - 1s/epoch - 4ms/step
Epoch 7/300
283/283 - 1s - loss: 1.9275 - accuracy: 0.4026 - val_loss: 1.8395 - val_accuracy: 0.4000 - 1s/epoch - 4ms/step
Epoch 8/300
283/283 - 1s - loss: 1.8391 - accuracy: 0.4199 - val_loss: 1.7678 - val_accuracy: 0.4210 - 1s/epoch - 4ms/step
Epoch 9/300
283/283 - 1s - loss: 1.7720 - accuracy: 0.4393 - val_loss: 1.6541 - val_accuracy: 0.4637 - 1s/epoch - 4ms/step
Epoch 10/300
283/283 - 1s - loss: 1.6887 - accuracy: 0.4658 - val_loss: 1.7132 - val_accuracy: 0.4350 - 1s/epoch - 4ms/step
Epoch 11/300
283/283 - 1s - loss: 1.6333 - accuracy: 0.4795 - val_loss: 1.6824 - val_accuracy: 0.4500 - 1s/epoch - 4ms/step
Epoch 12/300
283/283 - 1s - loss: 1.5633 - accuracy: 0.4966 - val_loss: 1.4250 - val_accuracy: 0.5480 - 1s/epoch - 4ms/step
Epoch 13/300
283/283 - 1s - loss: 1.5111 - accuracy: 0.5116 - val_loss: 1.3878 - val_accuracy: 0.5570 - 1s/epoch - 4ms/step
Epoch 14/300
283/283 - 1s - loss: 1.4515 - accuracy: 0.5364 - val_loss: 1.4584 - val_accuracy: 0.5327 - 1s/epoch - 4ms/step
Epoch 15/300
283/283 - 1s - loss: 1.4485 - accuracy: 0.5411 - val_loss: 1.3013 - val_accuracy: 0.5833 - 1s/epoch - 4ms/step
Epoch 16/300
283/283 - 1s - loss: 1.3917 - accuracy: 0.5503 - val_loss: 1.2499 - val_accuracy: 0.6227 - 1s/epoch - 4ms/step
Epoch 17/300
283/283 - 1s - loss: 1.3694 - accuracy: 0.5623 - val_loss: 1.2354 - val_accuracy: 0.5967 - 1s/epoch - 4ms/step
Epoch 18/300
283/283 - 1s - loss: 1.3375 - accuracy: 0.5714 - val_loss: 1.2662 - val_accuracy: 0.5993 - 1s/epoch - 4ms/step
Epoch 19/300
283/283 - 1s - loss: 1.3021 - accuracy: 0.5776 - val_loss: 1.2725 - val_accuracy: 0.5897 - 1s/epoch - 4ms/step
Epoch 20/300
283/283 - 1s - loss: 1.2703 - accuracy: 0.5846 - val_loss: 1.1179 - val_accuracy: 0.6453 - 1s/epoch - 4ms/step
Epoch 21/300
283/283 - 1s - loss: 1.2592 - accuracy: 0.5927 - val_loss: 1.1403 - val_accuracy: 0.6433 - 1s/epoch - 4ms/step
Epoch 22/300
283/283 - 1s - loss: 1.2724 - accuracy: 0.5857 - val_loss: 1.1366 - val_accuracy: 0.6450 - 1s/epoch - 4ms/step
Epoch 23/300
283/283 - 1s - loss: 1.2330 - accuracy: 0.6005 - val_loss: 1.1607 - val_accuracy: 0.6370 - 1s/epoch - 4ms/step
Epoch 24/300
283/283 - 1s - loss: 1.1852 - accuracy: 0.6101 - val_loss: 1.0979 - val_accuracy: 0.6563 - 1s/epoch - 4ms/step
Epoch 25/300
283/283 - 1s - loss: 1.1639 - accuracy: 0.6164 - val_loss: 1.0875 - val_accuracy: 0.6617 - 1s/epoch - 4ms/step
Epoch 26/300
283/283 - 1s - loss: 1.1705 - accuracy: 0.6111 - val_loss: 1.0990 - val_accuracy: 0.6587 - 1s/epoch - 4ms/step
Epoch 27/300
283/283 - 1s - loss: 1.1276 - accuracy: 0.6284 - val_loss: 1.0418 - val_accuracy: 0.6770 - 1s/epoch - 4ms/step
Epoch 28/300
283/283 - 1s - loss: 1.1303 - accuracy: 0.6296 - val_loss: 1.0353 - val_accuracy: 0.6797 - 1s/epoch - 4ms/step
Epoch 29/300
283/283 - 1s - loss: 1.0934 - accuracy: 0.6423 - val_loss: 1.0912 - val_accuracy: 0.6637 - 1s/epoch - 4ms/step
Epoch 30/300
283/283 - 1s - loss: 1.0921 - accuracy: 0.6392 - val_loss: 1.0040 - val_accuracy: 0.6890 - 1s/epoch - 4ms/step
Epoch 31/300
283/283 - 1s - loss: 1.0624 - accuracy: 0.6535 - val_loss: 1.0834 - val_accuracy: 0.6577 - 1s/epoch - 4ms/step
Epoch 32/300
283/283 - 1s - loss: 1.0614 - accuracy: 0.6520 - val_loss: 0.9979 - val_accuracy: 0.6837 - 1s/epoch - 4ms/step
Epoch 33/300
283/283 - 1s - loss: 1.0240 - accuracy: 0.6624 - val_loss: 0.9607 - val_accuracy: 0.7030 - 1s/epoch - 4ms/step
Epoch 34/300
283/283 - 1s - loss: 1.0126 - accuracy: 0.6607 - val_loss: 1.0190 - val_accuracy: 0.6913 - 1s/epoch - 4ms/step
Epoch 35/300
283/283 - 1s - loss: 1.0271 - accuracy: 0.6661 - val_loss: 1.0449 - val_accuracy: 0.6840 - 1s/epoch - 4ms/step
Epoch 36/300
283/283 - 1s - loss: 1.0016 - accuracy: 0.6638 - val_loss: 1.0300 - val_accuracy: 0.6767 - 1s/epoch - 4ms/step
Epoch 37/300
283/283 - 1s - loss: 0.9630 - accuracy: 0.6807 - val_loss: 1.1201 - val_accuracy: 0.6643 - 1s/epoch - 4ms/step
Epoch 38/300
283/283 - 1s - loss: 0.9567 - accuracy: 0.6794 - val_loss: 1.0279 - val_accuracy: 0.6857 - 1s/epoch - 4ms/step
Epoch 39/300
283/283 - 1s - loss: 0.9528 - accuracy: 0.6848 - val_loss: 1.0361 - val_accuracy: 0.6853 - 1s/epoch - 4ms/step
Epoch 40/300
283/283 - 1s - loss: 0.9329 - accuracy: 0.6897 - val_loss: 1.1854 - val_accuracy: 0.6520 - 1s/epoch - 4ms/step
Epoch 41/300
283/283 - 1s - loss: 0.9408 - accuracy: 0.6869 - val_loss: 0.9601 - val_accuracy: 0.7037 - 1s/epoch - 4ms/step
Epoch 42/300
283/283 - 1s - loss: 0.9164 - accuracy: 0.6937 - val_loss: 0.9108 - val_accuracy: 0.7283 - 1s/epoch - 4ms/step
Epoch 43/300
283/283 - 1s - loss: 0.9281 - accuracy: 0.6951 - val_loss: 0.9048 - val_accuracy: 0.7320 - 1s/epoch - 5ms/step
Epoch 44/300
283/283 - 1s - loss: 0.8990 - accuracy: 0.7039 - val_loss: 1.0257 - val_accuracy: 0.6913 - 1s/epoch - 4ms/step
Epoch 45/300
283/283 - 1s - loss: 0.8816 - accuracy: 0.7080 - val_loss: 0.9517 - val_accuracy: 0.7160 - 1s/epoch - 4ms/step
Epoch 46/300
283/283 - 1s - loss: 0.8914 - accuracy: 0.6989 - val_loss: 0.9446 - val_accuracy: 0.7253 - 1s/epoch - 4ms/step
Epoch 47/300
283/283 - 1s - loss: 0.8497 - accuracy: 0.7150 - val_loss: 0.8555 - val_accuracy: 0.7487 - 1s/epoch - 4ms/step
Epoch 48/300
283/283 - 1s - loss: 0.8242 - accuracy: 0.7214 - val_loss: 0.8457 - val_accuracy: 0.7503 - 1s/epoch - 4ms/step
Epoch 49/300
283/283 - 1s - loss: 0.8518 - accuracy: 0.7205 - val_loss: 0.8802 - val_accuracy: 0.7477 - 1s/epoch - 4ms/step
Epoch 50/300
283/283 - 1s - loss: 0.8067 - accuracy: 0.7280 - val_loss: 0.8763 - val_accuracy: 0.7470 - 1s/epoch - 4ms/step
Epoch 51/300
283/283 - 1s - loss: 0.8122 - accuracy: 0.7293 - val_loss: 0.8521 - val_accuracy: 0.7550 - 1s/epoch - 4ms/step
Epoch 52/300
283/283 - 1s - loss: 0.7702 - accuracy: 0.7411 - val_loss: 1.0250 - val_accuracy: 0.6993 - 1s/epoch - 4ms/step
Epoch 53/300
283/283 - 1s - loss: 0.8156 - accuracy: 0.7266 - val_loss: 0.8267 - val_accuracy: 0.7607 - 1s/epoch - 4ms/step
Epoch 54/300
283/283 - 1s - loss: 0.7676 - accuracy: 0.7432 - val_loss: 0.8517 - val_accuracy: 0.7473 - 1s/epoch - 4ms/step
Epoch 55/300
283/283 - 1s - loss: 0.7671 - accuracy: 0.7466 - val_loss: 0.8345 - val_accuracy: 0.7653 - 1s/epoch - 4ms/step
Epoch 56/300
283/283 - 1s - loss: 0.7632 - accuracy: 0.7512 - val_loss: 0.8147 - val_accuracy: 0.7593 - 1s/epoch - 4ms/step
Epoch 57/300
283/283 - 1s - loss: 0.7761 - accuracy: 0.7410 - val_loss: 0.8169 - val_accuracy: 0.7650 - 1s/epoch - 4ms/step
Epoch 58/300
283/283 - 1s - loss: 0.7628 - accuracy: 0.7453 - val_loss: 0.8814 - val_accuracy: 0.7400 - 1s/epoch - 5ms/step
Epoch 59/300
283/283 - 1s - loss: 0.7283 - accuracy: 0.7551 - val_loss: 0.7834 - val_accuracy: 0.7730 - 1s/epoch - 4ms/step
Epoch 60/300
283/283 - 1s - loss: 0.7267 - accuracy: 0.7579 - val_loss: 0.7850 - val_accuracy: 0.7760 - 1s/epoch - 4ms/step
Epoch 61/300
283/283 - 1s - loss: 0.7231 - accuracy: 0.7581 - val_loss: 0.9390 - val_accuracy: 0.7293 - 1s/epoch - 4ms/step
Epoch 62/300
283/283 - 1s - loss: 0.7389 - accuracy: 0.7568 - val_loss: 0.8118 - val_accuracy: 0.7700 - 1s/epoch - 4ms/step
Epoch 63/300
283/283 - 1s - loss: 0.7223 - accuracy: 0.7584 - val_loss: 0.8551 - val_accuracy: 0.7607 - 1s/epoch - 4ms/step
Epoch 64/300
283/283 - 1s - loss: 0.7129 - accuracy: 0.7628 - val_loss: 0.9686 - val_accuracy: 0.7383 - 1s/epoch - 4ms/step
Epoch 65/300
283/283 - 1s - loss: 0.6648 - accuracy: 0.7784 - val_loss: 0.8030 - val_accuracy: 0.7790 - 1s/epoch - 4ms/step
Epoch 66/300
283/283 - 1s - loss: 0.6962 - accuracy: 0.7703 - val_loss: 1.0090 - val_accuracy: 0.7280 - 1s/epoch - 4ms/step
Epoch 67/300
283/283 - 1s - loss: 0.7373 - accuracy: 0.7565 - val_loss: 0.9441 - val_accuracy: 0.7287 - 1s/epoch - 4ms/step
Epoch 68/300
283/283 - 1s - loss: 0.6657 - accuracy: 0.7764 - val_loss: 0.8653 - val_accuracy: 0.7677 - 1s/epoch - 4ms/step
Epoch 69/300
283/283 - 1s - loss: 0.6534 - accuracy: 0.7829 - val_loss: 0.8783 - val_accuracy: 0.7440 - 1s/epoch - 4ms/step
Epoch 70/300
283/283 - 1s - loss: 0.6903 - accuracy: 0.7650 - val_loss: 0.7778 - val_accuracy: 0.7823 - 1s/epoch - 4ms/step
Epoch 71/300
283/283 - 1s - loss: 0.6664 - accuracy: 0.7779 - val_loss: 0.8712 - val_accuracy: 0.7657 - 1s/epoch - 4ms/step
Epoch 72/300
283/283 - 1s - loss: 0.6343 - accuracy: 0.7832 - val_loss: 0.9044 - val_accuracy: 0.7627 - 1s/epoch - 4ms/step
Epoch 73/300
283/283 - 1s - loss: 0.6554 - accuracy: 0.7808 - val_loss: 0.8100 - val_accuracy: 0.7710 - 1s/epoch - 4ms/step
Epoch 74/300
283/283 - 1s - loss: 0.6298 - accuracy: 0.7900 - val_loss: 0.8039 - val_accuracy: 0.7870 - 1s/epoch - 4ms/step
Epoch 75/300
283/283 - 1s - loss: 0.6089 - accuracy: 0.7911 - val_loss: 0.9103 - val_accuracy: 0.7493 - 1s/epoch - 4ms/step
Epoch 76/300
283/283 - 1s - loss: 0.6242 - accuracy: 0.7967 - val_loss: 0.9285 - val_accuracy: 0.7460 - 1s/epoch - 4ms/step
Epoch 77/300
283/283 - 1s - loss: 0.6215 - accuracy: 0.7955 - val_loss: 0.8452 - val_accuracy: 0.7707 - 1s/epoch - 5ms/step
Epoch 78/300
283/283 - 1s - loss: 0.6253 - accuracy: 0.7911 - val_loss: 0.7904 - val_accuracy: 0.7827 - 1s/epoch - 4ms/step
Epoch 79/300
283/283 - 1s - loss: 0.6028 - accuracy: 0.8019 - val_loss: 0.7923 - val_accuracy: 0.7853 - 1s/epoch - 4ms/step
Epoch 80/300
283/283 - 1s - loss: 0.6119 - accuracy: 0.7981 - val_loss: 0.7997 - val_accuracy: 0.7797 - 1s/epoch - 4ms/step
Epoch 81/300
283/283 - 1s - loss: 0.6265 - accuracy: 0.7969 - val_loss: 0.7987 - val_accuracy: 0.7883 - 1s/epoch - 4ms/step
Epoch 82/300
283/283 - 1s - loss: 0.5458 - accuracy: 0.8199 - val_loss: 0.7891 - val_accuracy: 0.7887 - 1s/epoch - 5ms/step
Epoch 83/300
283/283 - 1s - loss: 0.5957 - accuracy: 0.8039 - val_loss: 0.9033 - val_accuracy: 0.7713 - 1s/epoch - 4ms/step
Epoch 84/300
283/283 - 1s - loss: 0.5774 - accuracy: 0.8090 - val_loss: 0.8252 - val_accuracy: 0.7800 - 1s/epoch - 4ms/step
Epoch 85/300
283/283 - 1s - loss: 0.5555 - accuracy: 0.8203 - val_loss: 0.9088 - val_accuracy: 0.7647 - 1s/epoch - 4ms/step
Epoch 86/300
283/283 - 1s - loss: 0.5969 - accuracy: 0.8016 - val_loss: 0.9000 - val_accuracy: 0.7770 - 1s/epoch - 4ms/step
Epoch 87/300
283/283 - 1s - loss: 0.5765 - accuracy: 0.8095 - val_loss: 0.8581 - val_accuracy: 0.7893 - 1s/epoch - 4ms/step
Epoch 88/300
283/283 - 1s - loss: 0.5542 - accuracy: 0.8166 - val_loss: 0.8227 - val_accuracy: 0.7920 - 1s/epoch - 4ms/step
Epoch 89/300
283/283 - 1s - loss: 0.5789 - accuracy: 0.8125 - val_loss: 0.7923 - val_accuracy: 0.7910 - 1s/epoch - 4ms/step
Epoch 90/300
283/283 - 1s - loss: 0.5193 - accuracy: 0.8281 - val_loss: 0.9567 - val_accuracy: 0.7560 - 1s/epoch - 4ms/step
Epoch 91/300
283/283 - 1s - loss: 0.5308 - accuracy: 0.8220 - val_loss: 0.7564 - val_accuracy: 0.7913 - 1s/epoch - 4ms/step
Epoch 92/300
283/283 - 1s - loss: 0.5584 - accuracy: 0.8223 - val_loss: 0.8272 - val_accuracy: 0.7893 - 1s/epoch - 4ms/step
Epoch 93/300
283/283 - 1s - loss: 0.5164 - accuracy: 0.8295 - val_loss: 0.7645 - val_accuracy: 0.7923 - 1s/epoch - 4ms/step
Epoch 94/300
283/283 - 1s - loss: 0.5482 - accuracy: 0.8200 - val_loss: 0.8663 - val_accuracy: 0.7807 - 1s/epoch - 4ms/step
Epoch 95/300
283/283 - 1s - loss: 0.5271 - accuracy: 0.8311 - val_loss: 0.9183 - val_accuracy: 0.7790 - 1s/epoch - 4ms/step
Epoch 96/300
283/283 - 1s - loss: 0.5320 - accuracy: 0.8265 - val_loss: 1.0680 - val_accuracy: 0.7627 - 1s/epoch - 4ms/step
Epoch 97/300
283/283 - 1s - loss: 0.5274 - accuracy: 0.8213 - val_loss: 0.8292 - val_accuracy: 0.7983 - 1s/epoch - 4ms/step
Epoch 98/300
283/283 - 1s - loss: 0.5129 - accuracy: 0.8286 - val_loss: 0.7916 - val_accuracy: 0.8013 - 1s/epoch - 4ms/step
Epoch 99/300
283/283 - 1s - loss: 0.5228 - accuracy: 0.8299 - val_loss: 0.8376 - val_accuracy: 0.7893 - 1s/epoch - 4ms/step
Epoch 100/300
283/283 - 1s - loss: 0.5041 - accuracy: 0.8322 - val_loss: 0.8607 - val_accuracy: 0.7827 - 1s/epoch - 4ms/step
Epoch 101/300
283/283 - 1s - loss: 0.4844 - accuracy: 0.8389 - val_loss: 0.9106 - val_accuracy: 0.7853 - 1s/epoch - 4ms/step
Epoch 102/300
283/283 - 1s - loss: 0.4888 - accuracy: 0.8367 - val_loss: 0.8914 - val_accuracy: 0.7853 - 1s/epoch - 4ms/step
Epoch 103/300
283/283 - 1s - loss: 0.4973 - accuracy: 0.8365 - val_loss: 0.8903 - val_accuracy: 0.7917 - 1s/epoch - 4ms/step
Epoch 104/300
283/283 - 1s - loss: 0.5103 - accuracy: 0.8323 - val_loss: 0.8535 - val_accuracy: 0.7963 - 1s/epoch - 4ms/step
Epoch 105/300
283/283 - 1s - loss: 0.4848 - accuracy: 0.8391 - val_loss: 0.8781 - val_accuracy: 0.7920 - 1s/epoch - 4ms/step
Epoch 106/300
283/283 - 1s - loss: 0.4705 - accuracy: 0.8487 - val_loss: 0.8640 - val_accuracy: 0.7953 - 1s/epoch - 4ms/step
Epoch 107/300
283/283 - 1s - loss: 0.5084 - accuracy: 0.8404 - val_loss: 0.9201 - val_accuracy: 0.7760 - 1s/epoch - 4ms/step
Epoch 108/300
283/283 - 1s - loss: 0.4837 - accuracy: 0.8496 - val_loss: 0.8853 - val_accuracy: 0.7833 - 1s/epoch - 4ms/step
Epoch 109/300
283/283 - 1s - loss: 0.4976 - accuracy: 0.8386 - val_loss: 0.8580 - val_accuracy: 0.7910 - 1s/epoch - 4ms/step
Epoch 110/300
283/283 - 1s - loss: 0.4864 - accuracy: 0.8432 - val_loss: 0.9199 - val_accuracy: 0.7877 - 1s/epoch - 4ms/step
Epoch 111/300
283/283 - 1s - loss: 0.4934 - accuracy: 0.8438 - val_loss: 1.0693 - val_accuracy: 0.7523 - 1s/epoch - 4ms/step
Epoch 112/300
283/283 - 1s - loss: 0.4454 - accuracy: 0.8572 - val_loss: 0.9064 - val_accuracy: 0.7833 - 1s/epoch - 4ms/step
Epoch 113/300
283/283 - 1s - loss: 0.4688 - accuracy: 0.8505 - val_loss: 0.8808 - val_accuracy: 0.7930 - 1s/epoch - 4ms/step
Epoch 114/300
283/283 - 1s - loss: 0.4452 - accuracy: 0.8525 - val_loss: 0.9373 - val_accuracy: 0.7793 - 1s/epoch - 4ms/step
Epoch 115/300
283/283 - 1s - loss: 0.4457 - accuracy: 0.8560 - val_loss: 1.1024 - val_accuracy: 0.7573 - 1s/epoch - 4ms/step
Epoch 116/300
283/283 - 1s - loss: 0.4901 - accuracy: 0.8419 - val_loss: 0.8895 - val_accuracy: 0.7970 - 1s/epoch - 4ms/step
Epoch 117/300
283/283 - 1s - loss: 0.4454 - accuracy: 0.8530 - val_loss: 0.8834 - val_accuracy: 0.7977 - 1s/epoch - 4ms/step
Epoch 118/300
283/283 - 1s - loss: 0.4604 - accuracy: 0.8546 - val_loss: 0.8400 - val_accuracy: 0.8040 - 1s/epoch - 4ms/step
Epoch 119/300
283/283 - 1s - loss: 0.4638 - accuracy: 0.8530 - val_loss: 0.9483 - val_accuracy: 0.7737 - 1s/epoch - 4ms/step
Epoch 120/300
283/283 - 1s - loss: 0.4327 - accuracy: 0.8611 - val_loss: 0.9507 - val_accuracy: 0.7943 - 1s/epoch - 4ms/step
Epoch 121/300
283/283 - 1s - loss: 0.4526 - accuracy: 0.8551 - val_loss: 0.9027 - val_accuracy: 0.7893 - 1s/epoch - 4ms/step
Epoch 122/300
283/283 - 1s - loss: 0.4547 - accuracy: 0.8558 - val_loss: 0.9184 - val_accuracy: 0.7920 - 1s/epoch - 4ms/step
Epoch 123/300
283/283 - 1s - loss: 0.4087 - accuracy: 0.8687 - val_loss: 0.9014 - val_accuracy: 0.7960 - 1s/epoch - 4ms/step
Epoch 124/300
283/283 - 1s - loss: 0.4014 - accuracy: 0.8714 - val_loss: 0.8588 - val_accuracy: 0.8123 - 1s/epoch - 4ms/step
Epoch 125/300
283/283 - 1s - loss: 0.4301 - accuracy: 0.8651 - val_loss: 1.0037 - val_accuracy: 0.7727 - 1s/epoch - 4ms/step
Epoch 126/300
283/283 - 1s - loss: 0.4038 - accuracy: 0.8710 - val_loss: 0.9286 - val_accuracy: 0.8093 - 1s/epoch - 5ms/step
Epoch 127/300
283/283 - 1s - loss: 0.4201 - accuracy: 0.8658 - val_loss: 1.0987 - val_accuracy: 0.7720 - 1s/epoch - 4ms/step
Epoch 128/300
283/283 - 1s - loss: 0.4297 - accuracy: 0.8658 - val_loss: 0.9787 - val_accuracy: 0.7827 - 1s/epoch - 4ms/step
Epoch 129/300
283/283 - 1s - loss: 0.4491 - accuracy: 0.8536 - val_loss: 0.9213 - val_accuracy: 0.7977 - 1s/epoch - 4ms/step
Epoch 130/300
283/283 - 1s - loss: 0.4203 - accuracy: 0.8628 - val_loss: 1.0757 - val_accuracy: 0.7643 - 1s/epoch - 4ms/step
Epoch 131/300
283/283 - 1s - loss: 0.4037 - accuracy: 0.8706 - val_loss: 0.9914 - val_accuracy: 0.7960 - 1s/epoch - 4ms/step
Epoch 132/300
283/283 - 1s - loss: 0.4131 - accuracy: 0.8656 - val_loss: 0.9210 - val_accuracy: 0.7977 - 1s/epoch - 4ms/step
Epoch 133/300
283/283 - 1s - loss: 0.4093 - accuracy: 0.8693 - val_loss: 0.9749 - val_accuracy: 0.7950 - 1s/epoch - 4ms/step
Epoch 134/300
283/283 - 1s - loss: 0.3880 - accuracy: 0.8755 - val_loss: 0.9718 - val_accuracy: 0.7960 - 1s/epoch - 4ms/step
Epoch 135/300
283/283 - 1s - loss: 0.3914 - accuracy: 0.8773 - val_loss: 0.9609 - val_accuracy: 0.8033 - 1s/epoch - 4ms/step
Epoch 136/300
283/283 - 1s - loss: 0.3910 - accuracy: 0.8785 - val_loss: 1.1285 - val_accuracy: 0.7783 - 1s/epoch - 4ms/step
Epoch 137/300
283/283 - 1s - loss: 0.4299 - accuracy: 0.8670 - val_loss: 0.9187 - val_accuracy: 0.7970 - 1s/epoch - 4ms/step
Epoch 138/300
283/283 - 1s - loss: 0.4047 - accuracy: 0.8714 - val_loss: 0.9399 - val_accuracy: 0.8090 - 1s/epoch - 4ms/step
Epoch 139/300
283/283 - 1s - loss: 0.3826 - accuracy: 0.8784 - val_loss: 1.0083 - val_accuracy: 0.7790 - 1s/epoch - 4ms/step
Epoch 140/300
283/283 - 1s - loss: 0.3941 - accuracy: 0.8765 - val_loss: 0.9963 - val_accuracy: 0.7887 - 1s/epoch - 4ms/step
Epoch 141/300
283/283 - 1s - loss: 0.3913 - accuracy: 0.8775 - val_loss: 1.0020 - val_accuracy: 0.7817 - 1s/epoch - 4ms/step
Epoch 142/300
283/283 - 1s - loss: 0.3683 - accuracy: 0.8806 - val_loss: 1.0068 - val_accuracy: 0.8013 - 1s/epoch - 4ms/step
Epoch 143/300
283/283 - 1s - loss: 0.3931 - accuracy: 0.8790 - val_loss: 0.9503 - val_accuracy: 0.7853 - 1s/epoch - 4ms/step
Epoch 144/300
283/283 - 1s - loss: 0.3928 - accuracy: 0.8775 - val_loss: 1.0095 - val_accuracy: 0.7867 - 1s/epoch - 4ms/step
Epoch 145/300
283/283 - 1s - loss: 0.3958 - accuracy: 0.8735 - val_loss: 0.9686 - val_accuracy: 0.7960 - 1s/epoch - 4ms/step
Epoch 146/300
283/283 - 1s - loss: 0.3738 - accuracy: 0.8858 - val_loss: 1.1791 - val_accuracy: 0.7733 - 1s/epoch - 4ms/step
Epoch 147/300
283/283 - 1s - loss: 0.3466 - accuracy: 0.8880 - val_loss: 1.0815 - val_accuracy: 0.7920 - 1s/epoch - 4ms/step
Epoch 148/300
283/283 - 1s - loss: 0.3727 - accuracy: 0.8825 - val_loss: 1.0038 - val_accuracy: 0.8000 - 1s/epoch - 4ms/step
Epoch 149/300
283/283 - 1s - loss: 0.3578 - accuracy: 0.8854 - val_loss: 1.0735 - val_accuracy: 0.7890 - 1s/epoch - 4ms/step
Epoch 150/300
283/283 - 1s - loss: 0.3548 - accuracy: 0.8898 - val_loss: 1.0359 - val_accuracy: 0.7933 - 1s/epoch - 4ms/step
Epoch 151/300
283/283 - 1s - loss: 0.3730 - accuracy: 0.8862 - val_loss: 0.9935 - val_accuracy: 0.7950 - 1s/epoch - 4ms/step
Epoch 152/300
283/283 - 1s - loss: 0.3605 - accuracy: 0.8857 - val_loss: 1.0255 - val_accuracy: 0.7997 - 1s/epoch - 4ms/step
Epoch 153/300
283/283 - 1s - loss: 0.3699 - accuracy: 0.8824 - val_loss: 1.0190 - val_accuracy: 0.7917 - 1s/epoch - 4ms/step
Epoch 154/300
283/283 - 1s - loss: 0.3771 - accuracy: 0.8827 - val_loss: 0.9451 - val_accuracy: 0.7913 - 1s/epoch - 4ms/step
Epoch 155/300
283/283 - 1s - loss: 0.3611 - accuracy: 0.8848 - val_loss: 0.9575 - val_accuracy: 0.7990 - 1s/epoch - 4ms/step
Epoch 156/300
283/283 - 1s - loss: 0.3613 - accuracy: 0.8850 - val_loss: 1.0040 - val_accuracy: 0.7993 - 1s/epoch - 4ms/step
Epoch 157/300
283/283 - 1s - loss: 0.3524 - accuracy: 0.8882 - val_loss: 1.0366 - val_accuracy: 0.7967 - 1s/epoch - 4ms/step
Epoch 158/300
283/283 - 1s - loss: 0.3660 - accuracy: 0.8880 - val_loss: 0.9877 - val_accuracy: 0.8057 - 1s/epoch - 4ms/step
Epoch 159/300
283/283 - 1s - loss: 0.3403 - accuracy: 0.8942 - val_loss: 1.0786 - val_accuracy: 0.7927 - 1s/epoch - 4ms/step
Epoch 160/300
283/283 - 1s - loss: 0.4417 - accuracy: 0.8686 - val_loss: 1.1056 - val_accuracy: 0.7697 - 1s/epoch - 4ms/step
Epoch 161/300
283/283 - 1s - loss: 0.3418 - accuracy: 0.8932 - val_loss: 1.0011 - val_accuracy: 0.7970 - 1s/epoch - 4ms/step
Epoch 162/300
283/283 - 1s - loss: 0.3371 - accuracy: 0.8958 - val_loss: 1.0302 - val_accuracy: 0.7933 - 1s/epoch - 5ms/step
Epoch 163/300
283/283 - 1s - loss: 0.3452 - accuracy: 0.8930 - val_loss: 1.0098 - val_accuracy: 0.8047 - 1s/epoch - 4ms/step
Epoch 164/300
283/283 - 1s - loss: 0.3253 - accuracy: 0.9006 - val_loss: 1.0711 - val_accuracy: 0.7923 - 1s/epoch - 4ms/step
Epoch 165/300
283/283 - 1s - loss: 0.3316 - accuracy: 0.8967 - val_loss: 0.9604 - val_accuracy: 0.7993 - 1s/epoch - 4ms/step
Epoch 166/300
283/283 - 1s - loss: 0.3338 - accuracy: 0.8926 - val_loss: 1.1320 - val_accuracy: 0.7917 - 1s/epoch - 4ms/step
Epoch 167/300
283/283 - 1s - loss: 0.4011 - accuracy: 0.8788 - val_loss: 1.0862 - val_accuracy: 0.7840 - 1s/epoch - 4ms/step
Epoch 168/300
283/283 - 1s - loss: 0.3426 - accuracy: 0.8922 - val_loss: 1.1471 - val_accuracy: 0.7860 - 1s/epoch - 4ms/step
Epoch 169/300
283/283 - 1s - loss: 0.3243 - accuracy: 0.8973 - val_loss: 1.0822 - val_accuracy: 0.7993 - 1s/epoch - 4ms/step
Epoch 170/300
283/283 - 1s - loss: 0.3111 - accuracy: 0.9025 - val_loss: 1.0963 - val_accuracy: 0.7937 - 1s/epoch - 4ms/step
Epoch 171/300
283/283 - 1s - loss: 0.3660 - accuracy: 0.8885 - val_loss: 1.0567 - val_accuracy: 0.7997 - 1s/epoch - 4ms/step
Epoch 172/300
283/283 - 1s - loss: 0.3162 - accuracy: 0.9055 - val_loss: 1.1083 - val_accuracy: 0.7973 - 1s/epoch - 4ms/step
Epoch 173/300
283/283 - 1s - loss: 0.3170 - accuracy: 0.9026 - val_loss: 1.0414 - val_accuracy: 0.7927 - 1s/epoch - 4ms/step
Epoch 174/300
283/283 - 1s - loss: 0.3232 - accuracy: 0.9047 - val_loss: 1.2663 - val_accuracy: 0.7780 - 1s/epoch - 4ms/step
Epoch 175/300
283/283 - 1s - loss: 0.3357 - accuracy: 0.8982 - val_loss: 1.1731 - val_accuracy: 0.7893 - 1s/epoch - 4ms/step
Epoch 176/300
283/283 - 1s - loss: 0.3288 - accuracy: 0.8999 - val_loss: 1.1959 - val_accuracy: 0.7807 - 1s/epoch - 4ms/step
Epoch 177/300
283/283 - 1s - loss: 0.3205 - accuracy: 0.8994 - val_loss: 1.0934 - val_accuracy: 0.8017 - 1s/epoch - 5ms/step
Epoch 178/300
283/283 - 1s - loss: 0.3054 - accuracy: 0.9074 - val_loss: 1.0826 - val_accuracy: 0.7997 - 1s/epoch - 4ms/step
Epoch 179/300
283/283 - 1s - loss: 0.3229 - accuracy: 0.9014 - val_loss: 1.0334 - val_accuracy: 0.7957 - 1s/epoch - 4ms/step
Epoch 180/300
283/283 - 1s - loss: 0.3258 - accuracy: 0.8993 - val_loss: 1.0944 - val_accuracy: 0.8027 - 1s/epoch - 4ms/step
Epoch 181/300
283/283 - 1s - loss: 0.2942 - accuracy: 0.9129 - val_loss: 1.1599 - val_accuracy: 0.8007 - 1s/epoch - 4ms/step
Epoch 182/300
283/283 - 1s - loss: 0.3023 - accuracy: 0.9108 - val_loss: 1.1662 - val_accuracy: 0.7860 - 1s/epoch - 4ms/step
Epoch 183/300
283/283 - 1s - loss: 0.3160 - accuracy: 0.9056 - val_loss: 1.1346 - val_accuracy: 0.7910 - 1s/epoch - 4ms/step
Epoch 184/300
283/283 - 1s - loss: 0.2944 - accuracy: 0.9128 - val_loss: 1.0517 - val_accuracy: 0.8073 - 1s/epoch - 4ms/step
Epoch 185/300
283/283 - 1s - loss: 0.3179 - accuracy: 0.9041 - val_loss: 1.2973 - val_accuracy: 0.7793 - 1s/epoch - 4ms/step
Epoch 186/300
283/283 - 1s - loss: 0.3523 - accuracy: 0.8983 - val_loss: 1.1796 - val_accuracy: 0.7930 - 1s/epoch - 4ms/step
Epoch 187/300
283/283 - 1s - loss: 0.2953 - accuracy: 0.9104 - val_loss: 1.1777 - val_accuracy: 0.8010 - 1s/epoch - 4ms/step
Epoch 188/300
283/283 - 1s - loss: 0.3194 - accuracy: 0.9010 - val_loss: 1.1365 - val_accuracy: 0.7947 - 1s/epoch - 4ms/step
Epoch 189/300
283/283 - 1s - loss: 0.3367 - accuracy: 0.8999 - val_loss: 1.1697 - val_accuracy: 0.7900 - 1s/epoch - 4ms/step
Epoch 190/300
283/283 - 1s - loss: 0.2833 - accuracy: 0.9145 - val_loss: 1.1307 - val_accuracy: 0.8010 - 1s/epoch - 4ms/step
Epoch 191/300
283/283 - 1s - loss: 0.2904 - accuracy: 0.9137 - val_loss: 1.2340 - val_accuracy: 0.7810 - 1s/epoch - 4ms/step
Epoch 192/300
283/283 - 1s - loss: 0.3178 - accuracy: 0.9043 - val_loss: 1.1162 - val_accuracy: 0.7947 - 1s/epoch - 4ms/step
Epoch 193/300
283/283 - 1s - loss: 0.2861 - accuracy: 0.9157 - val_loss: 1.1517 - val_accuracy: 0.7983 - 1s/epoch - 4ms/step
Epoch 194/300
283/283 - 1s - loss: 0.2755 - accuracy: 0.9148 - val_loss: 1.3031 - val_accuracy: 0.7923 - 1s/epoch - 4ms/step
Epoch 195/300
283/283 - 1s - loss: 0.2928 - accuracy: 0.9101 - val_loss: 1.0559 - val_accuracy: 0.7983 - 1s/epoch - 4ms/step
Epoch 196/300
283/283 - 1s - loss: 0.3021 - accuracy: 0.9080 - val_loss: 1.1484 - val_accuracy: 0.7800 - 1s/epoch - 4ms/step
Epoch 197/300
283/283 - 1s - loss: 0.3093 - accuracy: 0.9046 - val_loss: 1.1288 - val_accuracy: 0.8010 - 1s/epoch - 4ms/step
Epoch 198/300
283/283 - 1s - loss: 0.2918 - accuracy: 0.9074 - val_loss: 1.4267 - val_accuracy: 0.7693 - 1s/epoch - 4ms/step
Epoch 199/300
283/283 - 1s - loss: 0.2741 - accuracy: 0.9188 - val_loss: 1.1940 - val_accuracy: 0.7983 - 1s/epoch - 4ms/step
Epoch 200/300
283/283 - 1s - loss: 0.2831 - accuracy: 0.9152 - val_loss: 1.1935 - val_accuracy: 0.7963 - 1s/epoch - 4ms/step
Epoch 201/300
283/283 - 1s - loss: 0.2805 - accuracy: 0.9175 - val_loss: 1.2605 - val_accuracy: 0.7927 - 1s/epoch - 4ms/step
Epoch 202/300
283/283 - 1s - loss: 0.2932 - accuracy: 0.9116 - val_loss: 1.3942 - val_accuracy: 0.7763 - 1s/epoch - 4ms/step
Epoch 203/300
283/283 - 1s - loss: 0.3472 - accuracy: 0.8920 - val_loss: 1.1404 - val_accuracy: 0.7960 - 1s/epoch - 4ms/step
Epoch 204/300
283/283 - 1s - loss: 0.2848 - accuracy: 0.9163 - val_loss: 1.1130 - val_accuracy: 0.7947 - 1s/epoch - 4ms/step
Epoch 205/300
283/283 - 1s - loss: 0.2672 - accuracy: 0.9189 - val_loss: 1.1109 - val_accuracy: 0.8010 - 1s/epoch - 4ms/step
Epoch 206/300
283/283 - 1s - loss: 0.2822 - accuracy: 0.9179 - val_loss: 1.1096 - val_accuracy: 0.7923 - 1s/epoch - 4ms/step
Epoch 207/300
283/283 - 1s - loss: 0.2645 - accuracy: 0.9209 - val_loss: 1.4944 - val_accuracy: 0.7627 - 1s/epoch - 4ms/step
Epoch 208/300
283/283 - 1s - loss: 0.2866 - accuracy: 0.9122 - val_loss: 1.2218 - val_accuracy: 0.7950 - 1s/epoch - 4ms/step
Epoch 209/300
283/283 - 1s - loss: 0.2927 - accuracy: 0.9145 - val_loss: 1.0955 - val_accuracy: 0.7960 - 1s/epoch - 4ms/step
Epoch 210/300
283/283 - 1s - loss: 0.2850 - accuracy: 0.9148 - val_loss: 1.1840 - val_accuracy: 0.7980 - 1s/epoch - 4ms/step
Epoch 211/300
283/283 - 1s - loss: 0.2934 - accuracy: 0.9166 - val_loss: 1.1442 - val_accuracy: 0.7987 - 1s/epoch - 4ms/step
Epoch 212/300
283/283 - 1s - loss: 0.2742 - accuracy: 0.9200 - val_loss: 1.4180 - val_accuracy: 0.7630 - 1s/epoch - 4ms/step
Epoch 213/300
283/283 - 1s - loss: 0.2571 - accuracy: 0.9251 - val_loss: 1.1914 - val_accuracy: 0.7983 - 1s/epoch - 4ms/step
Epoch 214/300
283/283 - 1s - loss: 0.2800 - accuracy: 0.9174 - val_loss: 1.2203 - val_accuracy: 0.7843 - 1s/epoch - 4ms/step
Epoch 215/300
283/283 - 1s - loss: 0.2533 - accuracy: 0.9248 - val_loss: 1.2254 - val_accuracy: 0.7977 - 1s/epoch - 4ms/step
Epoch 216/300
283/283 - 1s - loss: 0.2618 - accuracy: 0.9220 - val_loss: 1.4186 - val_accuracy: 0.7660 - 1s/epoch - 4ms/step
Epoch 217/300
283/283 - 1s - loss: 0.2873 - accuracy: 0.9128 - val_loss: 1.2321 - val_accuracy: 0.7930 - 1s/epoch - 4ms/step
Epoch 218/300
283/283 - 1s - loss: 0.2602 - accuracy: 0.9248 - val_loss: 1.3534 - val_accuracy: 0.7793 - 1s/epoch - 4ms/step
Epoch 219/300
283/283 - 1s - loss: 0.2621 - accuracy: 0.9209 - val_loss: 1.4190 - val_accuracy: 0.7760 - 1s/epoch - 4ms/step
Epoch 220/300
283/283 - 1s - loss: 0.2881 - accuracy: 0.9171 - val_loss: 1.3922 - val_accuracy: 0.7697 - 1s/epoch - 4ms/step
Epoch 221/300
283/283 - 1s - loss: 0.2739 - accuracy: 0.9187 - val_loss: 1.4558 - val_accuracy: 0.7697 - 1s/epoch - 4ms/step
Epoch 222/300
283/283 - 1s - loss: 0.2796 - accuracy: 0.9138 - val_loss: 1.2145 - val_accuracy: 0.7987 - 1s/epoch - 4ms/step
Epoch 223/300
283/283 - 1s - loss: 0.2755 - accuracy: 0.9173 - val_loss: 1.1463 - val_accuracy: 0.7897 - 1s/epoch - 4ms/step
Epoch 224/300
283/283 - 1s - loss: 0.2701 - accuracy: 0.9210 - val_loss: 1.1013 - val_accuracy: 0.7873 - 1s/epoch - 4ms/step
Epoch 225/300
283/283 - 1s - loss: 0.2607 - accuracy: 0.9242 - val_loss: 1.2183 - val_accuracy: 0.7980 - 1s/epoch - 4ms/step
Epoch 226/300
283/283 - 1s - loss: 0.2716 - accuracy: 0.9209 - val_loss: 1.2247 - val_accuracy: 0.7943 - 1s/epoch - 4ms/step
Epoch 227/300
283/283 - 1s - loss: 0.2756 - accuracy: 0.9202 - val_loss: 1.2362 - val_accuracy: 0.7970 - 1s/epoch - 5ms/step
Epoch 228/300
283/283 - 1s - loss: 0.2479 - accuracy: 0.9301 - val_loss: 1.4131 - val_accuracy: 0.7750 - 1s/epoch - 4ms/step
Epoch 229/300
283/283 - 1s - loss: 0.2931 - accuracy: 0.9159 - val_loss: 1.1607 - val_accuracy: 0.8013 - 1s/epoch - 4ms/step
Epoch 230/300
283/283 - 1s - loss: 0.2356 - accuracy: 0.9301 - val_loss: 1.5370 - val_accuracy: 0.7623 - 1s/epoch - 4ms/step
Epoch 231/300
283/283 - 1s - loss: 0.2699 - accuracy: 0.9212 - val_loss: 1.2875 - val_accuracy: 0.7833 - 1s/epoch - 4ms/step
Epoch 232/300
283/283 - 1s - loss: 0.2511 - accuracy: 0.9272 - val_loss: 1.3482 - val_accuracy: 0.7693 - 1s/epoch - 4ms/step
Epoch 233/300
283/283 - 1s - loss: 0.2641 - accuracy: 0.9217 - val_loss: 1.2619 - val_accuracy: 0.7863 - 1s/epoch - 4ms/step
Epoch 234/300
283/283 - 1s - loss: 0.2824 - accuracy: 0.9199 - val_loss: 1.4691 - val_accuracy: 0.7563 - 1s/epoch - 4ms/step
Epoch 235/300
283/283 - 1s - loss: 0.2542 - accuracy: 0.9273 - val_loss: 1.2564 - val_accuracy: 0.7880 - 1s/epoch - 4ms/step
Epoch 236/300
283/283 - 1s - loss: 0.2244 - accuracy: 0.9341 - val_loss: 1.3651 - val_accuracy: 0.7910 - 1s/epoch - 4ms/step
Epoch 237/300
283/283 - 1s - loss: 0.2514 - accuracy: 0.9250 - val_loss: 1.2797 - val_accuracy: 0.7860 - 1s/epoch - 4ms/step
Epoch 238/300
283/283 - 1s - loss: 0.2514 - accuracy: 0.9265 - val_loss: 1.4776 - val_accuracy: 0.7763 - 1s/epoch - 4ms/step
Epoch 239/300
283/283 - 1s - loss: 0.2608 - accuracy: 0.9271 - val_loss: 1.3656 - val_accuracy: 0.7837 - 1s/epoch - 4ms/step
Epoch 240/300
283/283 - 1s - loss: 0.2616 - accuracy: 0.9259 - val_loss: 1.2925 - val_accuracy: 0.7877 - 1s/epoch - 4ms/step
Epoch 241/300
283/283 - 1s - loss: 0.2922 - accuracy: 0.9136 - val_loss: 1.3443 - val_accuracy: 0.7933 - 1s/epoch - 4ms/step
Epoch 242/300
283/283 - 1s - loss: 0.2366 - accuracy: 0.9292 - val_loss: 1.3947 - val_accuracy: 0.7853 - 1s/epoch - 4ms/step
Epoch 243/300
283/283 - 1s - loss: 0.2341 - accuracy: 0.9338 - val_loss: 1.2420 - val_accuracy: 0.7903 - 1s/epoch - 4ms/step
Epoch 244/300
283/283 - 1s - loss: 0.2340 - accuracy: 0.9355 - val_loss: 1.2221 - val_accuracy: 0.7947 - 1s/epoch - 4ms/step
Epoch 245/300
283/283 - 1s - loss: 0.2300 - accuracy: 0.9300 - val_loss: 1.3198 - val_accuracy: 0.7917 - 1s/epoch - 4ms/step
Epoch 246/300
283/283 - 1s - loss: 0.2281 - accuracy: 0.9356 - val_loss: 1.3303 - val_accuracy: 0.7787 - 1s/epoch - 4ms/step
Epoch 247/300
283/283 - 1s - loss: 0.2820 - accuracy: 0.9210 - val_loss: 1.5014 - val_accuracy: 0.7430 - 1s/epoch - 4ms/step
Epoch 248/300
283/283 - 1s - loss: 0.3142 - accuracy: 0.9096 - val_loss: 1.2643 - val_accuracy: 0.7890 - 1s/epoch - 4ms/step
Epoch 249/300
283/283 - 1s - loss: 0.2232 - accuracy: 0.9366 - val_loss: 1.3945 - val_accuracy: 0.7850 - 1s/epoch - 4ms/step
Epoch 250/300
283/283 - 1s - loss: 0.2417 - accuracy: 0.9317 - val_loss: 1.1633 - val_accuracy: 0.8087 - 1s/epoch - 4ms/step
Epoch 251/300
283/283 - 1s - loss: 0.2471 - accuracy: 0.9266 - val_loss: 1.3094 - val_accuracy: 0.7933 - 1s/epoch - 4ms/step
Epoch 252/300
283/283 - 1s - loss: 0.2481 - accuracy: 0.9301 - val_loss: 1.2007 - val_accuracy: 0.7913 - 1s/epoch - 4ms/step
Epoch 253/300
283/283 - 1s - loss: 0.2323 - accuracy: 0.9321 - val_loss: 1.4629 - val_accuracy: 0.7833 - 1s/epoch - 4ms/step
Epoch 254/300
283/283 - 1s - loss: 0.2387 - accuracy: 0.9301 - val_loss: 1.1606 - val_accuracy: 0.7973 - 1s/epoch - 4ms/step
Epoch 255/300
283/283 - 1s - loss: 0.2372 - accuracy: 0.9309 - val_loss: 1.3615 - val_accuracy: 0.7903 - 1s/epoch - 4ms/step
Epoch 256/300
283/283 - 1s - loss: 0.2327 - accuracy: 0.9339 - val_loss: 1.4382 - val_accuracy: 0.7817 - 1s/epoch - 4ms/step
Epoch 257/300
283/283 - 1s - loss: 0.2281 - accuracy: 0.9363 - val_loss: 1.3481 - val_accuracy: 0.7843 - 1s/epoch - 4ms/step
Epoch 258/300
283/283 - 1s - loss: 0.2380 - accuracy: 0.9359 - val_loss: 1.4001 - val_accuracy: 0.7777 - 1s/epoch - 4ms/step
Epoch 259/300
283/283 - 1s - loss: 0.2575 - accuracy: 0.9265 - val_loss: 1.3470 - val_accuracy: 0.7790 - 1s/epoch - 4ms/step
Epoch 260/300
283/283 - 1s - loss: 0.2250 - accuracy: 0.9340 - val_loss: 1.2973 - val_accuracy: 0.8003 - 1s/epoch - 4ms/step
Epoch 261/300
283/283 - 1s - loss: 0.2142 - accuracy: 0.9401 - val_loss: 1.5075 - val_accuracy: 0.7830 - 1s/epoch - 4ms/step
Epoch 262/300
283/283 - 1s - loss: 0.2269 - accuracy: 0.9362 - val_loss: 1.2600 - val_accuracy: 0.8020 - 1s/epoch - 4ms/step
Epoch 263/300
283/283 - 1s - loss: 0.2014 - accuracy: 0.9430 - val_loss: 1.3622 - val_accuracy: 0.7970 - 1s/epoch - 4ms/step
Epoch 264/300
283/283 - 1s - loss: 0.2488 - accuracy: 0.9319 - val_loss: 1.3931 - val_accuracy: 0.7653 - 1s/epoch - 4ms/step
Epoch 265/300
283/283 - 1s - loss: 0.3065 - accuracy: 0.9127 - val_loss: 1.4978 - val_accuracy: 0.7777 - 1s/epoch - 4ms/step
Epoch 266/300
283/283 - 1s - loss: 0.2399 - accuracy: 0.9312 - val_loss: 1.3711 - val_accuracy: 0.7913 - 1s/epoch - 4ms/step
Epoch 267/300
283/283 - 1s - loss: 0.2254 - accuracy: 0.9384 - val_loss: 1.6844 - val_accuracy: 0.7587 - 1s/epoch - 4ms/step
Epoch 268/300
283/283 - 1s - loss: 0.2348 - accuracy: 0.9334 - val_loss: 1.4405 - val_accuracy: 0.7907 - 1s/epoch - 4ms/step
Epoch 269/300
283/283 - 1s - loss: 0.2065 - accuracy: 0.9448 - val_loss: 1.4914 - val_accuracy: 0.7837 - 1s/epoch - 4ms/step
Epoch 270/300
283/283 - 1s - loss: 0.2048 - accuracy: 0.9427 - val_loss: 1.4739 - val_accuracy: 0.7953 - 1s/epoch - 4ms/step
Epoch 271/300
283/283 - 1s - loss: 0.2261 - accuracy: 0.9366 - val_loss: 1.4983 - val_accuracy: 0.7660 - 1s/epoch - 4ms/step
Epoch 272/300
283/283 - 1s - loss: 0.2489 - accuracy: 0.9255 - val_loss: 1.3971 - val_accuracy: 0.7940 - 1s/epoch - 4ms/step
Epoch 273/300
283/283 - 1s - loss: 0.2570 - accuracy: 0.9294 - val_loss: 1.3070 - val_accuracy: 0.7837 - 1s/epoch - 4ms/step
Epoch 274/300
283/283 - 1s - loss: 0.2715 - accuracy: 0.9232 - val_loss: 1.5456 - val_accuracy: 0.7627 - 1s/epoch - 4ms/step
Epoch 275/300
283/283 - 1s - loss: 0.2110 - accuracy: 0.9421 - val_loss: 1.3800 - val_accuracy: 0.7927 - 1s/epoch - 4ms/step
Epoch 276/300
283/283 - 1s - loss: 0.2049 - accuracy: 0.9472 - val_loss: 1.4535 - val_accuracy: 0.7913 - 1s/epoch - 4ms/step
Epoch 277/300
283/283 - 1s - loss: 0.2410 - accuracy: 0.9314 - val_loss: 1.2585 - val_accuracy: 0.7903 - 1s/epoch - 4ms/step
Epoch 278/300
283/283 - 1s - loss: 0.2122 - accuracy: 0.9405 - val_loss: 1.4418 - val_accuracy: 0.7923 - 1s/epoch - 4ms/step
Epoch 279/300
283/283 - 1s - loss: 0.2168 - accuracy: 0.9415 - val_loss: 1.4003 - val_accuracy: 0.8003 - 1s/epoch - 4ms/step
Epoch 280/300
283/283 - 1s - loss: 0.2182 - accuracy: 0.9393 - val_loss: 1.4800 - val_accuracy: 0.7913 - 1s/epoch - 5ms/step
Epoch 281/300
283/283 - 1s - loss: 0.2121 - accuracy: 0.9425 - val_loss: 1.6331 - val_accuracy: 0.7670 - 1s/epoch - 4ms/step
Epoch 282/300
283/283 - 1s - loss: 0.2393 - accuracy: 0.9361 - val_loss: 1.3554 - val_accuracy: 0.7807 - 1s/epoch - 4ms/step
Epoch 283/300
283/283 - 1s - loss: 0.2375 - accuracy: 0.9348 - val_loss: 1.2665 - val_accuracy: 0.7997 - 1s/epoch - 4ms/step
Epoch 284/300
283/283 - 1s - loss: 0.2241 - accuracy: 0.9376 - val_loss: 1.3791 - val_accuracy: 0.7880 - 1s/epoch - 4ms/step
Epoch 285/300
283/283 - 1s - loss: 0.2161 - accuracy: 0.9377 - val_loss: 1.4283 - val_accuracy: 0.7860 - 1s/epoch - 4ms/step
Epoch 286/300
283/283 - 1s - loss: 0.2070 - accuracy: 0.9424 - val_loss: 1.3078 - val_accuracy: 0.7933 - 1s/epoch - 4ms/step
Epoch 287/300
283/283 - 1s - loss: 0.2293 - accuracy: 0.9372 - val_loss: 1.4155 - val_accuracy: 0.7837 - 1s/epoch - 4ms/step
Epoch 288/300
283/283 - 1s - loss: 0.2143 - accuracy: 0.9402 - val_loss: 1.5785 - val_accuracy: 0.7907 - 1s/epoch - 4ms/step
Epoch 289/300
283/283 - 1s - loss: 0.2111 - accuracy: 0.9406 - val_loss: 1.4920 - val_accuracy: 0.7820 - 1s/epoch - 4ms/step
Epoch 290/300
283/283 - 1s - loss: 0.2112 - accuracy: 0.9389 - val_loss: 1.4132 - val_accuracy: 0.7863 - 1s/epoch - 4ms/step
Epoch 291/300
283/283 - 1s - loss: 0.2288 - accuracy: 0.9389 - val_loss: 1.4595 - val_accuracy: 0.7853 - 1s/epoch - 4ms/step
Epoch 292/300
283/283 - 1s - loss: 0.2671 - accuracy: 0.9260 - val_loss: 1.5033 - val_accuracy: 0.7873 - 1s/epoch - 4ms/step
Epoch 293/300
283/283 - 1s - loss: 0.1844 - accuracy: 0.9496 - val_loss: 1.5035 - val_accuracy: 0.7830 - 1s/epoch - 4ms/step
Epoch 294/300
283/283 - 1s - loss: 0.2185 - accuracy: 0.9394 - val_loss: 1.5891 - val_accuracy: 0.7500 - 1s/epoch - 4ms/step
Epoch 295/300
283/283 - 1s - loss: 0.2321 - accuracy: 0.9360 - val_loss: 1.4772 - val_accuracy: 0.7820 - 1s/epoch - 4ms/step
Epoch 296/300
283/283 - 1s - loss: 0.2049 - accuracy: 0.9447 - val_loss: 1.5046 - val_accuracy: 0.7887 - 1s/epoch - 4ms/step
Epoch 297/300
283/283 - 1s - loss: 0.2021 - accuracy: 0.9463 - val_loss: 1.5621 - val_accuracy: 0.7780 - 1s/epoch - 4ms/step
Epoch 298/300
283/283 - 1s - loss: 0.2016 - accuracy: 0.9468 - val_loss: 1.4417 - val_accuracy: 0.7997 - 1s/epoch - 4ms/step
Epoch 299/300
283/283 - 1s - loss: 0.2527 - accuracy: 0.9318 - val_loss: 1.4879 - val_accuracy: 0.7970 - 1s/epoch - 4ms/step
Epoch 300/300
283/283 - 1s - loss: 0.1939 - accuracy: 0.9465 - val_loss: 1.4999 - val_accuracy: 0.7990 - 1s/epoch - 4ms/step
94/94 - 0s - loss: 1.4999 - accuracy: 0.7990 - 201ms/epoch - 2ms/step
Baseline Error: 20.10%

We can clearly see that the model stops learning and basically completely flattens out past around the 100th epoch, it is wise to either lower the learning rate of the model altogether or implement a learning rate scheduler in order to improve the performance of the model.

#try lr scheduler

def lr_schedule(epoch):
    initial_lr = 0.001 #default of adam
    decay_factor = 0.8
    epochs_drop = 10
    new_lr = initial_lr * (decay_factor ** (epoch // epochs_drop))
    return new_lr

lr_scheduler = LearningRateScheduler(lr_schedule)

model_31new = models.Sequential([
    layers.Conv2D(32, (3, 3), activation='relu', input_shape=(31, 31, 1)),
    layers.MaxPooling2D((2, 2)),
    
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.MaxPooling2D((2, 2)),
    
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.MaxPooling2D((2, 2)),
    
    layers.Flatten(),
    
    layers.Dense(64, activation='relu', kernel_regularizer=regularizers.l2(0.001)),
    layers.Dropout(0.5), 
    
    layers.Dense(15, activation='softmax')
])


model_31new.compile(optimizer='adam',
                loss= 'categorical_crossentropy',
                metrics=['accuracy'])
    
history = model_31new.fit(X_train31, train_labels, validation_data=(X_val31, validation_labels),
                          epochs=300, batch_size=32, verbose=2, class_weight=class_weight,
                          callbacks=[lr_scheduler])

scores = model_31new.evaluate(X_val31, validation_labels, verbose=2)
print("Baseline Error: %.2f%%" % (100-scores[1]*100))
plot_history(history)
df31.loc[len(df31)] = ['LearningRateDecayModel', history.history['accuracy'][-1], history.history['val_accuracy'][-1], history.history['loss'][-1], history.history['val_loss'][-1], history]

Epoch 1/300
283/283 - 3s - loss: 3.1262 - accuracy: 0.1203 - val_loss: 2.8285 - val_accuracy: 0.1013 - lr: 0.0010 - 3s/epoch - 11ms/step
Epoch 2/300
283/283 - 1s - loss: 2.7520 - accuracy: 0.1954 - val_loss: 2.4456 - val_accuracy: 0.1913 - lr: 0.0010 - 1s/epoch - 4ms/step
Epoch 3/300
283/283 - 1s - loss: 2.4816 - accuracy: 0.2521 - val_loss: 2.2994 - val_accuracy: 0.2377 - lr: 0.0010 - 1s/epoch - 4ms/step
Epoch 4/300
283/283 - 1s - loss: 2.2671 - accuracy: 0.3042 - val_loss: 2.0560 - val_accuracy: 0.3273 - lr: 0.0010 - 1s/epoch - 4ms/step
Epoch 5/300
283/283 - 1s - loss: 2.1091 - accuracy: 0.3388 - val_loss: 1.9396 - val_accuracy: 0.3613 - lr: 0.0010 - 1s/epoch - 4ms/step
Epoch 6/300
283/283 - 1s - loss: 2.0190 - accuracy: 0.3658 - val_loss: 1.9670 - val_accuracy: 0.3650 - lr: 0.0010 - 1s/epoch - 4ms/step
Epoch 7/300
283/283 - 1s - loss: 1.9097 - accuracy: 0.3961 - val_loss: 1.8025 - val_accuracy: 0.3957 - lr: 0.0010 - 1s/epoch - 4ms/step
Epoch 8/300
283/283 - 1s - loss: 1.8434 - accuracy: 0.4201 - val_loss: 1.7814 - val_accuracy: 0.4053 - lr: 0.0010 - 1s/epoch - 4ms/step
Epoch 9/300
283/283 - 1s - loss: 1.7674 - accuracy: 0.4372 - val_loss: 1.8203 - val_accuracy: 0.4170 - lr: 0.0010 - 1s/epoch - 4ms/step
Epoch 10/300
283/283 - 1s - loss: 1.7073 - accuracy: 0.4608 - val_loss: 1.5570 - val_accuracy: 0.4907 - lr: 0.0010 - 1s/epoch - 4ms/step
Epoch 11/300
283/283 - 1s - loss: 1.6102 - accuracy: 0.4760 - val_loss: 1.4891 - val_accuracy: 0.5227 - lr: 8.0000e-04 - 1s/epoch - 4ms/step
Epoch 12/300
283/283 - 1s - loss: 1.5752 - accuracy: 0.4884 - val_loss: 1.4478 - val_accuracy: 0.5253 - lr: 8.0000e-04 - 1s/epoch - 4ms/step
Epoch 13/300
283/283 - 1s - loss: 1.5160 - accuracy: 0.5122 - val_loss: 1.3675 - val_accuracy: 0.5613 - lr: 8.0000e-04 - 1s/epoch - 4ms/step
Epoch 14/300
283/283 - 1s - loss: 1.5088 - accuracy: 0.5144 - val_loss: 1.4156 - val_accuracy: 0.5607 - lr: 8.0000e-04 - 1s/epoch - 4ms/step
Epoch 15/300
283/283 - 1s - loss: 1.4611 - accuracy: 0.5308 - val_loss: 1.4035 - val_accuracy: 0.5413 - lr: 8.0000e-04 - 1s/epoch - 4ms/step
Epoch 16/300
283/283 - 1s - loss: 1.4154 - accuracy: 0.5393 - val_loss: 1.3792 - val_accuracy: 0.5550 - lr: 8.0000e-04 - 1s/epoch - 4ms/step
Epoch 17/300
283/283 - 1s - loss: 1.3879 - accuracy: 0.5457 - val_loss: 1.2724 - val_accuracy: 0.5910 - lr: 8.0000e-04 - 1s/epoch - 4ms/step
Epoch 18/300
283/283 - 1s - loss: 1.3554 - accuracy: 0.5600 - val_loss: 1.2767 - val_accuracy: 0.5823 - lr: 8.0000e-04 - 1s/epoch - 4ms/step
Epoch 19/300
283/283 - 1s - loss: 1.3238 - accuracy: 0.5651 - val_loss: 1.2085 - val_accuracy: 0.6110 - lr: 8.0000e-04 - 1s/epoch - 4ms/step
Epoch 20/300
283/283 - 1s - loss: 1.3134 - accuracy: 0.5683 - val_loss: 1.1897 - val_accuracy: 0.6010 - lr: 8.0000e-04 - 1s/epoch - 4ms/step
Epoch 21/300
283/283 - 1s - loss: 1.2724 - accuracy: 0.5807 - val_loss: 1.1938 - val_accuracy: 0.6070 - lr: 6.4000e-04 - 1s/epoch - 4ms/step
Epoch 22/300
283/283 - 1s - loss: 1.2618 - accuracy: 0.5851 - val_loss: 1.1338 - val_accuracy: 0.6370 - lr: 6.4000e-04 - 1s/epoch - 4ms/step
Epoch 23/300
283/283 - 1s - loss: 1.2093 - accuracy: 0.6020 - val_loss: 1.1791 - val_accuracy: 0.6203 - lr: 6.4000e-04 - 1s/epoch - 4ms/step
Epoch 24/300
283/283 - 1s - loss: 1.2154 - accuracy: 0.6023 - val_loss: 1.0927 - val_accuracy: 0.6487 - lr: 6.4000e-04 - 1s/epoch - 4ms/step
Epoch 25/300
283/283 - 1s - loss: 1.1767 - accuracy: 0.6151 - val_loss: 1.0704 - val_accuracy: 0.6620 - lr: 6.4000e-04 - 1s/epoch - 4ms/step
Epoch 26/300
283/283 - 1s - loss: 1.1437 - accuracy: 0.6164 - val_loss: 1.1386 - val_accuracy: 0.6387 - lr: 6.4000e-04 - 1s/epoch - 4ms/step
Epoch 27/300
283/283 - 1s - loss: 1.1682 - accuracy: 0.6153 - val_loss: 1.2084 - val_accuracy: 0.6213 - lr: 6.4000e-04 - 1s/epoch - 4ms/step
Epoch 28/300
283/283 - 1s - loss: 1.1541 - accuracy: 0.6179 - val_loss: 1.0421 - val_accuracy: 0.6687 - lr: 6.4000e-04 - 1s/epoch - 4ms/step
Epoch 29/300
283/283 - 1s - loss: 1.1182 - accuracy: 0.6257 - val_loss: 1.1686 - val_accuracy: 0.6317 - lr: 6.4000e-04 - 1s/epoch - 5ms/step
Epoch 30/300
283/283 - 1s - loss: 1.1149 - accuracy: 0.6270 - val_loss: 1.1002 - val_accuracy: 0.6447 - lr: 6.4000e-04 - 1s/epoch - 5ms/step
Epoch 31/300
283/283 - 1s - loss: 1.0759 - accuracy: 0.6388 - val_loss: 1.0610 - val_accuracy: 0.6683 - lr: 5.1200e-04 - 1s/epoch - 4ms/step
Epoch 32/300
283/283 - 1s - loss: 1.0744 - accuracy: 0.6400 - val_loss: 1.1071 - val_accuracy: 0.6510 - lr: 5.1200e-04 - 1s/epoch - 4ms/step
Epoch 33/300
283/283 - 1s - loss: 1.0651 - accuracy: 0.6490 - val_loss: 0.9978 - val_accuracy: 0.6863 - lr: 5.1200e-04 - 1s/epoch - 4ms/step
Epoch 34/300
283/283 - 1s - loss: 1.0553 - accuracy: 0.6492 - val_loss: 1.0409 - val_accuracy: 0.6663 - lr: 5.1200e-04 - 1s/epoch - 4ms/step
Epoch 35/300
283/283 - 1s - loss: 1.0477 - accuracy: 0.6533 - val_loss: 0.9910 - val_accuracy: 0.6873 - lr: 5.1200e-04 - 1s/epoch - 4ms/step
Epoch 36/300
283/283 - 1s - loss: 1.0201 - accuracy: 0.6576 - val_loss: 0.9862 - val_accuracy: 0.6867 - lr: 5.1200e-04 - 1s/epoch - 4ms/step
Epoch 37/300
283/283 - 1s - loss: 1.0305 - accuracy: 0.6577 - val_loss: 1.0031 - val_accuracy: 0.6833 - lr: 5.1200e-04 - 1s/epoch - 4ms/step
Epoch 38/300
283/283 - 1s - loss: 0.9966 - accuracy: 0.6669 - val_loss: 1.0698 - val_accuracy: 0.6690 - lr: 5.1200e-04 - 1s/epoch - 4ms/step
Epoch 39/300
283/283 - 1s - loss: 0.9981 - accuracy: 0.6661 - val_loss: 0.9631 - val_accuracy: 0.7000 - lr: 5.1200e-04 - 1s/epoch - 4ms/step
Epoch 40/300
283/283 - 1s - loss: 0.9859 - accuracy: 0.6711 - val_loss: 1.0407 - val_accuracy: 0.6727 - lr: 5.1200e-04 - 1s/epoch - 4ms/step
Epoch 41/300
283/283 - 1s - loss: 0.9519 - accuracy: 0.6765 - val_loss: 0.9758 - val_accuracy: 0.6930 - lr: 4.0960e-04 - 1s/epoch - 4ms/step
Epoch 42/300
283/283 - 1s - loss: 0.9337 - accuracy: 0.6841 - val_loss: 1.0135 - val_accuracy: 0.6833 - lr: 4.0960e-04 - 1s/epoch - 4ms/step
Epoch 43/300
283/283 - 1s - loss: 0.9454 - accuracy: 0.6815 - val_loss: 0.9429 - val_accuracy: 0.7113 - lr: 4.0960e-04 - 1s/epoch - 4ms/step
Epoch 44/300
283/283 - 1s - loss: 0.9498 - accuracy: 0.6778 - val_loss: 0.9800 - val_accuracy: 0.6923 - lr: 4.0960e-04 - 1s/epoch - 4ms/step
Epoch 45/300
283/283 - 1s - loss: 0.9292 - accuracy: 0.6831 - val_loss: 0.9825 - val_accuracy: 0.6937 - lr: 4.0960e-04 - 1s/epoch - 4ms/step
Epoch 46/300
283/283 - 1s - loss: 0.9357 - accuracy: 0.6810 - val_loss: 0.9067 - val_accuracy: 0.7137 - lr: 4.0960e-04 - 1s/epoch - 4ms/step
Epoch 47/300
283/283 - 1s - loss: 0.8997 - accuracy: 0.6877 - val_loss: 0.9716 - val_accuracy: 0.7017 - lr: 4.0960e-04 - 1s/epoch - 4ms/step
Epoch 48/300
283/283 - 1s - loss: 0.9119 - accuracy: 0.6924 - val_loss: 0.9313 - val_accuracy: 0.7097 - lr: 4.0960e-04 - 1s/epoch - 4ms/step
Epoch 49/300
283/283 - 1s - loss: 0.8961 - accuracy: 0.6982 - val_loss: 0.9830 - val_accuracy: 0.7043 - lr: 4.0960e-04 - 1s/epoch - 4ms/step
Epoch 50/300
283/283 - 1s - loss: 0.8786 - accuracy: 0.6955 - val_loss: 1.0134 - val_accuracy: 0.6947 - lr: 4.0960e-04 - 1s/epoch - 4ms/step
Epoch 51/300
283/283 - 1s - loss: 0.8559 - accuracy: 0.7054 - val_loss: 0.9076 - val_accuracy: 0.7243 - lr: 3.2768e-04 - 1s/epoch - 4ms/step
Epoch 52/300
283/283 - 1s - loss: 0.8579 - accuracy: 0.7074 - val_loss: 0.8980 - val_accuracy: 0.7213 - lr: 3.2768e-04 - 1s/epoch - 4ms/step
Epoch 53/300
283/283 - 1s - loss: 0.8676 - accuracy: 0.7044 - val_loss: 0.8940 - val_accuracy: 0.7293 - lr: 3.2768e-04 - 1s/epoch - 4ms/step
Epoch 54/300
283/283 - 1s - loss: 0.8690 - accuracy: 0.7043 - val_loss: 0.9082 - val_accuracy: 0.7173 - lr: 3.2768e-04 - 1s/epoch - 4ms/step
Epoch 55/300
283/283 - 1s - loss: 0.8432 - accuracy: 0.7082 - val_loss: 0.9318 - val_accuracy: 0.7183 - lr: 3.2768e-04 - 1s/epoch - 4ms/step
Epoch 56/300
283/283 - 1s - loss: 0.8467 - accuracy: 0.7103 - val_loss: 0.9698 - val_accuracy: 0.7090 - lr: 3.2768e-04 - 1s/epoch - 4ms/step
Epoch 57/300
283/283 - 1s - loss: 0.8302 - accuracy: 0.7175 - val_loss: 0.9555 - val_accuracy: 0.7133 - lr: 3.2768e-04 - 1s/epoch - 4ms/step
Epoch 58/300
283/283 - 1s - loss: 0.8391 - accuracy: 0.7117 - val_loss: 0.9670 - val_accuracy: 0.7033 - lr: 3.2768e-04 - 1s/epoch - 4ms/step
Epoch 59/300
283/283 - 1s - loss: 0.8249 - accuracy: 0.7126 - val_loss: 0.9110 - val_accuracy: 0.7247 - lr: 3.2768e-04 - 1s/epoch - 4ms/step
Epoch 60/300
283/283 - 1s - loss: 0.8282 - accuracy: 0.7189 - val_loss: 0.9228 - val_accuracy: 0.7243 - lr: 3.2768e-04 - 1s/epoch - 4ms/step
Epoch 61/300
283/283 - 1s - loss: 0.8170 - accuracy: 0.7220 - val_loss: 0.8991 - val_accuracy: 0.7320 - lr: 2.6214e-04 - 1s/epoch - 4ms/step
Epoch 62/300
283/283 - 1s - loss: 0.8038 - accuracy: 0.7202 - val_loss: 0.9298 - val_accuracy: 0.7260 - lr: 2.6214e-04 - 1s/epoch - 4ms/step
Epoch 63/300
283/283 - 1s - loss: 0.8110 - accuracy: 0.7205 - val_loss: 0.8978 - val_accuracy: 0.7310 - lr: 2.6214e-04 - 1s/epoch - 4ms/step
Epoch 64/300
283/283 - 1s - loss: 0.7965 - accuracy: 0.7272 - val_loss: 0.8924 - val_accuracy: 0.7287 - lr: 2.6214e-04 - 1s/epoch - 4ms/step
Epoch 65/300
283/283 - 1s - loss: 0.7776 - accuracy: 0.7301 - val_loss: 0.8942 - val_accuracy: 0.7397 - lr: 2.6214e-04 - 1s/epoch - 4ms/step
Epoch 66/300
283/283 - 1s - loss: 0.7934 - accuracy: 0.7264 - val_loss: 0.9054 - val_accuracy: 0.7290 - lr: 2.6214e-04 - 1s/epoch - 4ms/step
Epoch 67/300
283/283 - 1s - loss: 0.7787 - accuracy: 0.7342 - val_loss: 0.8894 - val_accuracy: 0.7360 - lr: 2.6214e-04 - 1s/epoch - 4ms/step
Epoch 68/300
283/283 - 1s - loss: 0.7944 - accuracy: 0.7260 - val_loss: 0.9075 - val_accuracy: 0.7347 - lr: 2.6214e-04 - 1s/epoch - 4ms/step
Epoch 69/300
283/283 - 1s - loss: 0.7826 - accuracy: 0.7281 - val_loss: 0.9049 - val_accuracy: 0.7377 - lr: 2.6214e-04 - 1s/epoch - 4ms/step
Epoch 70/300
283/283 - 1s - loss: 0.7961 - accuracy: 0.7262 - val_loss: 0.9088 - val_accuracy: 0.7310 - lr: 2.6214e-04 - 1s/epoch - 4ms/step
Epoch 71/300
283/283 - 1s - loss: 0.7512 - accuracy: 0.7374 - val_loss: 0.8849 - val_accuracy: 0.7387 - lr: 2.0972e-04 - 1s/epoch - 4ms/step
Epoch 72/300
283/283 - 1s - loss: 0.7471 - accuracy: 0.7446 - val_loss: 0.8757 - val_accuracy: 0.7427 - lr: 2.0972e-04 - 1s/epoch - 4ms/step
Epoch 73/300
283/283 - 1s - loss: 0.7457 - accuracy: 0.7418 - val_loss: 0.9002 - val_accuracy: 0.7293 - lr: 2.0972e-04 - 1s/epoch - 4ms/step
Epoch 74/300
283/283 - 1s - loss: 0.7565 - accuracy: 0.7327 - val_loss: 0.8542 - val_accuracy: 0.7453 - lr: 2.0972e-04 - 1s/epoch - 4ms/step
Epoch 75/300
283/283 - 1s - loss: 0.7500 - accuracy: 0.7432 - val_loss: 0.8801 - val_accuracy: 0.7400 - lr: 2.0972e-04 - 1s/epoch - 4ms/step
Epoch 76/300
283/283 - 1s - loss: 0.7535 - accuracy: 0.7366 - val_loss: 0.9296 - val_accuracy: 0.7297 - lr: 2.0972e-04 - 1s/epoch - 5ms/step
Epoch 77/300
283/283 - 1s - loss: 0.7480 - accuracy: 0.7399 - val_loss: 0.9334 - val_accuracy: 0.7307 - lr: 2.0972e-04 - 1s/epoch - 4ms/step
Epoch 78/300
283/283 - 1s - loss: 0.7472 - accuracy: 0.7396 - val_loss: 0.8630 - val_accuracy: 0.7443 - lr: 2.0972e-04 - 1s/epoch - 4ms/step
Epoch 79/300
283/283 - 1s - loss: 0.7367 - accuracy: 0.7459 - val_loss: 0.8894 - val_accuracy: 0.7393 - lr: 2.0972e-04 - 1s/epoch - 4ms/step
Epoch 80/300
283/283 - 1s - loss: 0.7234 - accuracy: 0.7497 - val_loss: 0.8701 - val_accuracy: 0.7453 - lr: 2.0972e-04 - 1s/epoch - 4ms/step
Epoch 81/300
283/283 - 1s - loss: 0.7253 - accuracy: 0.7452 - val_loss: 0.8834 - val_accuracy: 0.7420 - lr: 1.6777e-04 - 1s/epoch - 4ms/step
Epoch 82/300
283/283 - 1s - loss: 0.7195 - accuracy: 0.7487 - val_loss: 0.9345 - val_accuracy: 0.7263 - lr: 1.6777e-04 - 1s/epoch - 4ms/step
Epoch 83/300
283/283 - 1s - loss: 0.7140 - accuracy: 0.7480 - val_loss: 0.8841 - val_accuracy: 0.7483 - lr: 1.6777e-04 - 1s/epoch - 4ms/step
Epoch 84/300
283/283 - 1s - loss: 0.7314 - accuracy: 0.7450 - val_loss: 0.8675 - val_accuracy: 0.7480 - lr: 1.6777e-04 - 1s/epoch - 4ms/step
Epoch 85/300
283/283 - 1s - loss: 0.7192 - accuracy: 0.7527 - val_loss: 0.8897 - val_accuracy: 0.7420 - lr: 1.6777e-04 - 1s/epoch - 4ms/step
Epoch 86/300
283/283 - 1s - loss: 0.7224 - accuracy: 0.7492 - val_loss: 0.8603 - val_accuracy: 0.7487 - lr: 1.6777e-04 - 1s/epoch - 4ms/step
Epoch 87/300
283/283 - 1s - loss: 0.7237 - accuracy: 0.7455 - val_loss: 0.8768 - val_accuracy: 0.7503 - lr: 1.6777e-04 - 1s/epoch - 4ms/step
Epoch 88/300
283/283 - 1s - loss: 0.7029 - accuracy: 0.7591 - val_loss: 0.8919 - val_accuracy: 0.7413 - lr: 1.6777e-04 - 1s/epoch - 4ms/step
Epoch 89/300
283/283 - 1s - loss: 0.7186 - accuracy: 0.7428 - val_loss: 0.9010 - val_accuracy: 0.7407 - lr: 1.6777e-04 - 1s/epoch - 4ms/step
Epoch 90/300
283/283 - 1s - loss: 0.6969 - accuracy: 0.7545 - val_loss: 0.8847 - val_accuracy: 0.7457 - lr: 1.6777e-04 - 1s/epoch - 4ms/step
Epoch 91/300
283/283 - 1s - loss: 0.6935 - accuracy: 0.7555 - val_loss: 0.9366 - val_accuracy: 0.7390 - lr: 1.3422e-04 - 1s/epoch - 4ms/step
Epoch 92/300
283/283 - 1s - loss: 0.6867 - accuracy: 0.7627 - val_loss: 0.8943 - val_accuracy: 0.7447 - lr: 1.3422e-04 - 1s/epoch - 4ms/step
Epoch 93/300
283/283 - 1s - loss: 0.7013 - accuracy: 0.7533 - val_loss: 0.9000 - val_accuracy: 0.7530 - lr: 1.3422e-04 - 1s/epoch - 4ms/step
Epoch 94/300
283/283 - 1s - loss: 0.6989 - accuracy: 0.7605 - val_loss: 0.8869 - val_accuracy: 0.7477 - lr: 1.3422e-04 - 1s/epoch - 4ms/step
Epoch 95/300
283/283 - 1s - loss: 0.6950 - accuracy: 0.7586 - val_loss: 0.8592 - val_accuracy: 0.7530 - lr: 1.3422e-04 - 1s/epoch - 4ms/step
Epoch 96/300
283/283 - 1s - loss: 0.6842 - accuracy: 0.7620 - val_loss: 0.9051 - val_accuracy: 0.7483 - lr: 1.3422e-04 - 1s/epoch - 4ms/step
Epoch 97/300
283/283 - 1s - loss: 0.6771 - accuracy: 0.7607 - val_loss: 0.9001 - val_accuracy: 0.7477 - lr: 1.3422e-04 - 1s/epoch - 4ms/step
Epoch 98/300
283/283 - 1s - loss: 0.6728 - accuracy: 0.7656 - val_loss: 0.8863 - val_accuracy: 0.7530 - lr: 1.3422e-04 - 1s/epoch - 4ms/step
Epoch 99/300
283/283 - 1s - loss: 0.6866 - accuracy: 0.7621 - val_loss: 0.8832 - val_accuracy: 0.7493 - lr: 1.3422e-04 - 1s/epoch - 4ms/step
Epoch 100/300
283/283 - 1s - loss: 0.6739 - accuracy: 0.7655 - val_loss: 0.8887 - val_accuracy: 0.7470 - lr: 1.3422e-04 - 1s/epoch - 4ms/step
Epoch 101/300
283/283 - 1s - loss: 0.6644 - accuracy: 0.7656 - val_loss: 0.9022 - val_accuracy: 0.7507 - lr: 1.0737e-04 - 1s/epoch - 4ms/step
Epoch 102/300
283/283 - 1s - loss: 0.6778 - accuracy: 0.7636 - val_loss: 0.8725 - val_accuracy: 0.7533 - lr: 1.0737e-04 - 1s/epoch - 4ms/step
Epoch 103/300
283/283 - 1s - loss: 0.6579 - accuracy: 0.7674 - val_loss: 0.8940 - val_accuracy: 0.7557 - lr: 1.0737e-04 - 1s/epoch - 4ms/step
Epoch 104/300
283/283 - 1s - loss: 0.6925 - accuracy: 0.7563 - val_loss: 0.8806 - val_accuracy: 0.7513 - lr: 1.0737e-04 - 1s/epoch - 4ms/step
Epoch 105/300
283/283 - 1s - loss: 0.6623 - accuracy: 0.7687 - val_loss: 0.9102 - val_accuracy: 0.7473 - lr: 1.0737e-04 - 1s/epoch - 4ms/step
Epoch 106/300
283/283 - 1s - loss: 0.6648 - accuracy: 0.7671 - val_loss: 0.8614 - val_accuracy: 0.7550 - lr: 1.0737e-04 - 1s/epoch - 4ms/step
Epoch 107/300
283/283 - 1s - loss: 0.6609 - accuracy: 0.7720 - val_loss: 0.8816 - val_accuracy: 0.7550 - lr: 1.0737e-04 - 1s/epoch - 4ms/step
Epoch 108/300
283/283 - 1s - loss: 0.6712 - accuracy: 0.7687 - val_loss: 0.8857 - val_accuracy: 0.7573 - lr: 1.0737e-04 - 1s/epoch - 4ms/step
Epoch 109/300
283/283 - 1s - loss: 0.6723 - accuracy: 0.7650 - val_loss: 0.8914 - val_accuracy: 0.7537 - lr: 1.0737e-04 - 1s/epoch - 4ms/step
Epoch 110/300
283/283 - 1s - loss: 0.6712 - accuracy: 0.7653 - val_loss: 0.9171 - val_accuracy: 0.7500 - lr: 1.0737e-04 - 1s/epoch - 4ms/step
Epoch 111/300
283/283 - 1s - loss: 0.6652 - accuracy: 0.7648 - val_loss: 0.8705 - val_accuracy: 0.7587 - lr: 8.5899e-05 - 1s/epoch - 4ms/step
Epoch 112/300
283/283 - 1s - loss: 0.6585 - accuracy: 0.7695 - val_loss: 0.8981 - val_accuracy: 0.7553 - lr: 8.5899e-05 - 1s/epoch - 4ms/step
Epoch 113/300
283/283 - 1s - loss: 0.6582 - accuracy: 0.7678 - val_loss: 0.8849 - val_accuracy: 0.7563 - lr: 8.5899e-05 - 1s/epoch - 4ms/step
Epoch 114/300
283/283 - 1s - loss: 0.6536 - accuracy: 0.7679 - val_loss: 0.8992 - val_accuracy: 0.7530 - lr: 8.5899e-05 - 1s/epoch - 4ms/step
Epoch 115/300
283/283 - 1s - loss: 0.6452 - accuracy: 0.7768 - val_loss: 0.9003 - val_accuracy: 0.7520 - lr: 8.5899e-05 - 1s/epoch - 4ms/step
Epoch 116/300
283/283 - 1s - loss: 0.6428 - accuracy: 0.7743 - val_loss: 0.8728 - val_accuracy: 0.7617 - lr: 8.5899e-05 - 1s/epoch - 4ms/step
Epoch 117/300
283/283 - 1s - loss: 0.6430 - accuracy: 0.7725 - val_loss: 0.8787 - val_accuracy: 0.7583 - lr: 8.5899e-05 - 1s/epoch - 4ms/step
Epoch 118/300
283/283 - 1s - loss: 0.6525 - accuracy: 0.7736 - val_loss: 0.8786 - val_accuracy: 0.7553 - lr: 8.5899e-05 - 1s/epoch - 4ms/step
Epoch 119/300
283/283 - 1s - loss: 0.6453 - accuracy: 0.7685 - val_loss: 0.8797 - val_accuracy: 0.7570 - lr: 8.5899e-05 - 1s/epoch - 4ms/step
Epoch 120/300
283/283 - 1s - loss: 0.6469 - accuracy: 0.7740 - val_loss: 0.8791 - val_accuracy: 0.7597 - lr: 8.5899e-05 - 1s/epoch - 4ms/step
Epoch 121/300
283/283 - 1s - loss: 0.6409 - accuracy: 0.7727 - val_loss: 0.8851 - val_accuracy: 0.7583 - lr: 6.8719e-05 - 1s/epoch - 4ms/step
Epoch 122/300
283/283 - 1s - loss: 0.6378 - accuracy: 0.7718 - val_loss: 0.8880 - val_accuracy: 0.7517 - lr: 6.8719e-05 - 1s/epoch - 4ms/step
Epoch 123/300
283/283 - 1s - loss: 0.6239 - accuracy: 0.7753 - val_loss: 0.8931 - val_accuracy: 0.7540 - lr: 6.8719e-05 - 1s/epoch - 4ms/step
Epoch 124/300
283/283 - 1s - loss: 0.6487 - accuracy: 0.7736 - val_loss: 0.8846 - val_accuracy: 0.7590 - lr: 6.8719e-05 - 1s/epoch - 4ms/step
Epoch 125/300
283/283 - 1s - loss: 0.6386 - accuracy: 0.7733 - val_loss: 0.8910 - val_accuracy: 0.7597 - lr: 6.8719e-05 - 1s/epoch - 4ms/step
Epoch 126/300
283/283 - 1s - loss: 0.6421 - accuracy: 0.7743 - val_loss: 0.8778 - val_accuracy: 0.7580 - lr: 6.8719e-05 - 1s/epoch - 4ms/step
Epoch 127/300
283/283 - 1s - loss: 0.6451 - accuracy: 0.7740 - val_loss: 0.8965 - val_accuracy: 0.7567 - lr: 6.8719e-05 - 1s/epoch - 5ms/step
Epoch 128/300
283/283 - 1s - loss: 0.6446 - accuracy: 0.7780 - val_loss: 0.8806 - val_accuracy: 0.7587 - lr: 6.8719e-05 - 1s/epoch - 4ms/step
Epoch 129/300
283/283 - 1s - loss: 0.6302 - accuracy: 0.7716 - val_loss: 0.8832 - val_accuracy: 0.7567 - lr: 6.8719e-05 - 1s/epoch - 4ms/step
Epoch 130/300
283/283 - 1s - loss: 0.6305 - accuracy: 0.7743 - val_loss: 0.8987 - val_accuracy: 0.7587 - lr: 6.8719e-05 - 1s/epoch - 4ms/step
Epoch 131/300
283/283 - 1s - loss: 0.6144 - accuracy: 0.7828 - val_loss: 0.8702 - val_accuracy: 0.7640 - lr: 5.4976e-05 - 1s/epoch - 4ms/step
Epoch 132/300
283/283 - 1s - loss: 0.6274 - accuracy: 0.7818 - val_loss: 0.8938 - val_accuracy: 0.7557 - lr: 5.4976e-05 - 1s/epoch - 4ms/step
Epoch 133/300
283/283 - 1s - loss: 0.6287 - accuracy: 0.7806 - val_loss: 0.8969 - val_accuracy: 0.7597 - lr: 5.4976e-05 - 1s/epoch - 4ms/step
Epoch 134/300
283/283 - 1s - loss: 0.6209 - accuracy: 0.7778 - val_loss: 0.8982 - val_accuracy: 0.7540 - lr: 5.4976e-05 - 1s/epoch - 4ms/step
Epoch 135/300
283/283 - 1s - loss: 0.6341 - accuracy: 0.7775 - val_loss: 0.8882 - val_accuracy: 0.7590 - lr: 5.4976e-05 - 1s/epoch - 4ms/step
Epoch 136/300
283/283 - 1s - loss: 0.6292 - accuracy: 0.7771 - val_loss: 0.8871 - val_accuracy: 0.7603 - lr: 5.4976e-05 - 1s/epoch - 4ms/step
Epoch 137/300
283/283 - 1s - loss: 0.6217 - accuracy: 0.7822 - val_loss: 0.8869 - val_accuracy: 0.7597 - lr: 5.4976e-05 - 1s/epoch - 4ms/step
Epoch 138/300
283/283 - 1s - loss: 0.6258 - accuracy: 0.7788 - val_loss: 0.8977 - val_accuracy: 0.7587 - lr: 5.4976e-05 - 1s/epoch - 4ms/step
Epoch 139/300
283/283 - 1s - loss: 0.6246 - accuracy: 0.7792 - val_loss: 0.8713 - val_accuracy: 0.7637 - lr: 5.4976e-05 - 1s/epoch - 4ms/step
Epoch 140/300
283/283 - 1s - loss: 0.6211 - accuracy: 0.7813 - val_loss: 0.8937 - val_accuracy: 0.7597 - lr: 5.4976e-05 - 1s/epoch - 4ms/step
Epoch 141/300
283/283 - 1s - loss: 0.6169 - accuracy: 0.7843 - val_loss: 0.9039 - val_accuracy: 0.7567 - lr: 4.3980e-05 - 1s/epoch - 4ms/step
Epoch 142/300
283/283 - 1s - loss: 0.6297 - accuracy: 0.7789 - val_loss: 0.8965 - val_accuracy: 0.7580 - lr: 4.3980e-05 - 1s/epoch - 4ms/step
Epoch 143/300
283/283 - 1s - loss: 0.6181 - accuracy: 0.7850 - val_loss: 0.8959 - val_accuracy: 0.7593 - lr: 4.3980e-05 - 1s/epoch - 4ms/step
Epoch 144/300
283/283 - 1s - loss: 0.6246 - accuracy: 0.7818 - val_loss: 0.8868 - val_accuracy: 0.7580 - lr: 4.3980e-05 - 1s/epoch - 4ms/step
Epoch 145/300
283/283 - 1s - loss: 0.6113 - accuracy: 0.7843 - val_loss: 0.8985 - val_accuracy: 0.7600 - lr: 4.3980e-05 - 1s/epoch - 4ms/step
Epoch 146/300
283/283 - 1s - loss: 0.6199 - accuracy: 0.7797 - val_loss: 0.9074 - val_accuracy: 0.7523 - lr: 4.3980e-05 - 1s/epoch - 4ms/step
Epoch 147/300
283/283 - 1s - loss: 0.6006 - accuracy: 0.7832 - val_loss: 0.8993 - val_accuracy: 0.7597 - lr: 4.3980e-05 - 1s/epoch - 4ms/step
Epoch 148/300
283/283 - 1s - loss: 0.6080 - accuracy: 0.7847 - val_loss: 0.8973 - val_accuracy: 0.7583 - lr: 4.3980e-05 - 1s/epoch - 4ms/step
Epoch 149/300
283/283 - 1s - loss: 0.6108 - accuracy: 0.7811 - val_loss: 0.8921 - val_accuracy: 0.7587 - lr: 4.3980e-05 - 1s/epoch - 4ms/step
Epoch 150/300
283/283 - 1s - loss: 0.6108 - accuracy: 0.7875 - val_loss: 0.8732 - val_accuracy: 0.7623 - lr: 4.3980e-05 - 1s/epoch - 4ms/step
Epoch 151/300
283/283 - 1s - loss: 0.6182 - accuracy: 0.7823 - val_loss: 0.9034 - val_accuracy: 0.7563 - lr: 3.5184e-05 - 1s/epoch - 4ms/step
Epoch 152/300
283/283 - 1s - loss: 0.6114 - accuracy: 0.7844 - val_loss: 0.9017 - val_accuracy: 0.7557 - lr: 3.5184e-05 - 1s/epoch - 4ms/step
Epoch 153/300
283/283 - 1s - loss: 0.6148 - accuracy: 0.7821 - val_loss: 0.8934 - val_accuracy: 0.7590 - lr: 3.5184e-05 - 1s/epoch - 4ms/step
Epoch 154/300
283/283 - 1s - loss: 0.6000 - accuracy: 0.7825 - val_loss: 0.8952 - val_accuracy: 0.7597 - lr: 3.5184e-05 - 1s/epoch - 4ms/step
Epoch 155/300
283/283 - 1s - loss: 0.6173 - accuracy: 0.7812 - val_loss: 0.8941 - val_accuracy: 0.7630 - lr: 3.5184e-05 - 1s/epoch - 4ms/step
Epoch 156/300
283/283 - 1s - loss: 0.6213 - accuracy: 0.7839 - val_loss: 0.8686 - val_accuracy: 0.7647 - lr: 3.5184e-05 - 1s/epoch - 4ms/step
Epoch 157/300
283/283 - 1s - loss: 0.5949 - accuracy: 0.7908 - val_loss: 0.8970 - val_accuracy: 0.7563 - lr: 3.5184e-05 - 1s/epoch - 4ms/step
Epoch 158/300
283/283 - 1s - loss: 0.6071 - accuracy: 0.7941 - val_loss: 0.9025 - val_accuracy: 0.7570 - lr: 3.5184e-05 - 1s/epoch - 4ms/step
Epoch 159/300
283/283 - 1s - loss: 0.6042 - accuracy: 0.7820 - val_loss: 0.8829 - val_accuracy: 0.7593 - lr: 3.5184e-05 - 1s/epoch - 4ms/step
Epoch 160/300
283/283 - 1s - loss: 0.6154 - accuracy: 0.7806 - val_loss: 0.9066 - val_accuracy: 0.7573 - lr: 3.5184e-05 - 1s/epoch - 4ms/step
Epoch 161/300
283/283 - 1s - loss: 0.6107 - accuracy: 0.7822 - val_loss: 0.8867 - val_accuracy: 0.7610 - lr: 2.8147e-05 - 1s/epoch - 4ms/step
Epoch 162/300
283/283 - 1s - loss: 0.5950 - accuracy: 0.7858 - val_loss: 0.8991 - val_accuracy: 0.7577 - lr: 2.8147e-05 - 1s/epoch - 4ms/step
Epoch 163/300
283/283 - 1s - loss: 0.6162 - accuracy: 0.7809 - val_loss: 0.8818 - val_accuracy: 0.7627 - lr: 2.8147e-05 - 1s/epoch - 4ms/step
Epoch 164/300
283/283 - 1s - loss: 0.6020 - accuracy: 0.7871 - val_loss: 0.8875 - val_accuracy: 0.7617 - lr: 2.8147e-05 - 1s/epoch - 4ms/step
Epoch 165/300
283/283 - 1s - loss: 0.6000 - accuracy: 0.7913 - val_loss: 0.8972 - val_accuracy: 0.7603 - lr: 2.8147e-05 - 1s/epoch - 4ms/step
Epoch 166/300
283/283 - 1s - loss: 0.6036 - accuracy: 0.7877 - val_loss: 0.8881 - val_accuracy: 0.7597 - lr: 2.8147e-05 - 1s/epoch - 4ms/step
Epoch 167/300
283/283 - 1s - loss: 0.5868 - accuracy: 0.7930 - val_loss: 0.9032 - val_accuracy: 0.7567 - lr: 2.8147e-05 - 1s/epoch - 4ms/step
Epoch 168/300
283/283 - 1s - loss: 0.6058 - accuracy: 0.7859 - val_loss: 0.8823 - val_accuracy: 0.7623 - lr: 2.8147e-05 - 1s/epoch - 4ms/step
Epoch 169/300
283/283 - 1s - loss: 0.6042 - accuracy: 0.7847 - val_loss: 0.8924 - val_accuracy: 0.7623 - lr: 2.8147e-05 - 1s/epoch - 4ms/step
Epoch 170/300
283/283 - 1s - loss: 0.6085 - accuracy: 0.7849 - val_loss: 0.9077 - val_accuracy: 0.7567 - lr: 2.8147e-05 - 1s/epoch - 4ms/step
Epoch 171/300
283/283 - 1s - loss: 0.6008 - accuracy: 0.7909 - val_loss: 0.8937 - val_accuracy: 0.7617 - lr: 2.2518e-05 - 1s/epoch - 4ms/step
Epoch 172/300
283/283 - 1s - loss: 0.5921 - accuracy: 0.7936 - val_loss: 0.8955 - val_accuracy: 0.7590 - lr: 2.2518e-05 - 1s/epoch - 4ms/step
Epoch 173/300
283/283 - 1s - loss: 0.5852 - accuracy: 0.7970 - val_loss: 0.9056 - val_accuracy: 0.7590 - lr: 2.2518e-05 - 1s/epoch - 4ms/step
Epoch 174/300
283/283 - 1s - loss: 0.5997 - accuracy: 0.7920 - val_loss: 0.8867 - val_accuracy: 0.7613 - lr: 2.2518e-05 - 1s/epoch - 4ms/step
Epoch 175/300
283/283 - 1s - loss: 0.6057 - accuracy: 0.7895 - val_loss: 0.9011 - val_accuracy: 0.7587 - lr: 2.2518e-05 - 1s/epoch - 4ms/step
Epoch 176/300
283/283 - 1s - loss: 0.5886 - accuracy: 0.7880 - val_loss: 0.9002 - val_accuracy: 0.7600 - lr: 2.2518e-05 - 1s/epoch - 4ms/step
Epoch 177/300
283/283 - 1s - loss: 0.6055 - accuracy: 0.7826 - val_loss: 0.8951 - val_accuracy: 0.7630 - lr: 2.2518e-05 - 1s/epoch - 4ms/step
Epoch 178/300
283/283 - 1s - loss: 0.5954 - accuracy: 0.7877 - val_loss: 0.9010 - val_accuracy: 0.7600 - lr: 2.2518e-05 - 1s/epoch - 4ms/step
Epoch 179/300
283/283 - 1s - loss: 0.5960 - accuracy: 0.7868 - val_loss: 0.8921 - val_accuracy: 0.7623 - lr: 2.2518e-05 - 1s/epoch - 4ms/step
Epoch 180/300
283/283 - 1s - loss: 0.6008 - accuracy: 0.7920 - val_loss: 0.9004 - val_accuracy: 0.7597 - lr: 2.2518e-05 - 1s/epoch - 4ms/step
Epoch 181/300
283/283 - 1s - loss: 0.5972 - accuracy: 0.7887 - val_loss: 0.8961 - val_accuracy: 0.7607 - lr: 1.8014e-05 - 1s/epoch - 4ms/step
Epoch 182/300
283/283 - 1s - loss: 0.5899 - accuracy: 0.7945 - val_loss: 0.8909 - val_accuracy: 0.7620 - lr: 1.8014e-05 - 1s/epoch - 4ms/step
Epoch 183/300
283/283 - 1s - loss: 0.6057 - accuracy: 0.7887 - val_loss: 0.9042 - val_accuracy: 0.7597 - lr: 1.8014e-05 - 1s/epoch - 4ms/step
Epoch 184/300
283/283 - 1s - loss: 0.5889 - accuracy: 0.7908 - val_loss: 0.9026 - val_accuracy: 0.7580 - lr: 1.8014e-05 - 1s/epoch - 4ms/step
Epoch 185/300
283/283 - 1s - loss: 0.5966 - accuracy: 0.7867 - val_loss: 0.9096 - val_accuracy: 0.7600 - lr: 1.8014e-05 - 1s/epoch - 4ms/step
Epoch 186/300
283/283 - 1s - loss: 0.5902 - accuracy: 0.7924 - val_loss: 0.8922 - val_accuracy: 0.7627 - lr: 1.8014e-05 - 1s/epoch - 4ms/step
Epoch 187/300
283/283 - 1s - loss: 0.5950 - accuracy: 0.7935 - val_loss: 0.8979 - val_accuracy: 0.7613 - lr: 1.8014e-05 - 1s/epoch - 4ms/step
Epoch 188/300
283/283 - 1s - loss: 0.5884 - accuracy: 0.7916 - val_loss: 0.8888 - val_accuracy: 0.7643 - lr: 1.8014e-05 - 1s/epoch - 4ms/step
Epoch 189/300
283/283 - 1s - loss: 0.5922 - accuracy: 0.7864 - val_loss: 0.8961 - val_accuracy: 0.7630 - lr: 1.8014e-05 - 1s/epoch - 4ms/step
Epoch 190/300
283/283 - 1s - loss: 0.6104 - accuracy: 0.7816 - val_loss: 0.9134 - val_accuracy: 0.7597 - lr: 1.8014e-05 - 1s/epoch - 4ms/step
Epoch 191/300
283/283 - 1s - loss: 0.5990 - accuracy: 0.7919 - val_loss: 0.8921 - val_accuracy: 0.7627 - lr: 1.4412e-05 - 1s/epoch - 4ms/step
Epoch 192/300
283/283 - 1s - loss: 0.5907 - accuracy: 0.7928 - val_loss: 0.8937 - val_accuracy: 0.7620 - lr: 1.4412e-05 - 1s/epoch - 4ms/step
Epoch 193/300
283/283 - 1s - loss: 0.5920 - accuracy: 0.7901 - val_loss: 0.9015 - val_accuracy: 0.7610 - lr: 1.4412e-05 - 1s/epoch - 4ms/step
Epoch 194/300
283/283 - 1s - loss: 0.5904 - accuracy: 0.7957 - val_loss: 0.8988 - val_accuracy: 0.7637 - lr: 1.4412e-05 - 1s/epoch - 4ms/step
Epoch 195/300
283/283 - 1s - loss: 0.5948 - accuracy: 0.7884 - val_loss: 0.8992 - val_accuracy: 0.7607 - lr: 1.4412e-05 - 1s/epoch - 4ms/step
Epoch 196/300
283/283 - 1s - loss: 0.5829 - accuracy: 0.7900 - val_loss: 0.8968 - val_accuracy: 0.7603 - lr: 1.4412e-05 - 1s/epoch - 4ms/step
Epoch 197/300
283/283 - 1s - loss: 0.5874 - accuracy: 0.7944 - val_loss: 0.8958 - val_accuracy: 0.7620 - lr: 1.4412e-05 - 1s/epoch - 4ms/step
Epoch 198/300
283/283 - 1s - loss: 0.5955 - accuracy: 0.7905 - val_loss: 0.8925 - val_accuracy: 0.7613 - lr: 1.4412e-05 - 1s/epoch - 4ms/step
Epoch 199/300
283/283 - 1s - loss: 0.5895 - accuracy: 0.7923 - val_loss: 0.8986 - val_accuracy: 0.7620 - lr: 1.4412e-05 - 1s/epoch - 4ms/step
Epoch 200/300
283/283 - 1s - loss: 0.6009 - accuracy: 0.7846 - val_loss: 0.9024 - val_accuracy: 0.7610 - lr: 1.4412e-05 - 1s/epoch - 4ms/step
Epoch 201/300
283/283 - 1s - loss: 0.5915 - accuracy: 0.7880 - val_loss: 0.8997 - val_accuracy: 0.7610 - lr: 1.1529e-05 - 1s/epoch - 4ms/step
Epoch 202/300
283/283 - 1s - loss: 0.5915 - accuracy: 0.7910 - val_loss: 0.8923 - val_accuracy: 0.7627 - lr: 1.1529e-05 - 1s/epoch - 4ms/step
Epoch 203/300
283/283 - 1s - loss: 0.5967 - accuracy: 0.7930 - val_loss: 0.9090 - val_accuracy: 0.7603 - lr: 1.1529e-05 - 1s/epoch - 4ms/step
Epoch 204/300
283/283 - 1s - loss: 0.5890 - accuracy: 0.7899 - val_loss: 0.9107 - val_accuracy: 0.7610 - lr: 1.1529e-05 - 1s/epoch - 4ms/step
Epoch 205/300
283/283 - 1s - loss: 0.5898 - accuracy: 0.7858 - val_loss: 0.8959 - val_accuracy: 0.7633 - lr: 1.1529e-05 - 1s/epoch - 4ms/step
Epoch 206/300
283/283 - 1s - loss: 0.6012 - accuracy: 0.7867 - val_loss: 0.9025 - val_accuracy: 0.7627 - lr: 1.1529e-05 - 1s/epoch - 4ms/step
Epoch 207/300
283/283 - 1s - loss: 0.5883 - accuracy: 0.7944 - val_loss: 0.9047 - val_accuracy: 0.7620 - lr: 1.1529e-05 - 1s/epoch - 4ms/step
Epoch 208/300
283/283 - 1s - loss: 0.5959 - accuracy: 0.7907 - val_loss: 0.9025 - val_accuracy: 0.7630 - lr: 1.1529e-05 - 1s/epoch - 4ms/step
Epoch 209/300
283/283 - 1s - loss: 0.5921 - accuracy: 0.7933 - val_loss: 0.9014 - val_accuracy: 0.7610 - lr: 1.1529e-05 - 1s/epoch - 4ms/step
Epoch 210/300
283/283 - 1s - loss: 0.5882 - accuracy: 0.7928 - val_loss: 0.8948 - val_accuracy: 0.7637 - lr: 1.1529e-05 - 1s/epoch - 4ms/step
Epoch 211/300
283/283 - 1s - loss: 0.5898 - accuracy: 0.7952 - val_loss: 0.9024 - val_accuracy: 0.7613 - lr: 9.2234e-06 - 1s/epoch - 4ms/step
Epoch 212/300
283/283 - 1s - loss: 0.5913 - accuracy: 0.7897 - val_loss: 0.8947 - val_accuracy: 0.7627 - lr: 9.2234e-06 - 1s/epoch - 4ms/step
Epoch 213/300
283/283 - 1s - loss: 0.5896 - accuracy: 0.7944 - val_loss: 0.8981 - val_accuracy: 0.7620 - lr: 9.2234e-06 - 1s/epoch - 4ms/step
Epoch 214/300
283/283 - 1s - loss: 0.5818 - accuracy: 0.7911 - val_loss: 0.8958 - val_accuracy: 0.7630 - lr: 9.2234e-06 - 1s/epoch - 4ms/step
Epoch 215/300
283/283 - 1s - loss: 0.5916 - accuracy: 0.7959 - val_loss: 0.8951 - val_accuracy: 0.7623 - lr: 9.2234e-06 - 1s/epoch - 4ms/step
Epoch 216/300
283/283 - 1s - loss: 0.5773 - accuracy: 0.7916 - val_loss: 0.8998 - val_accuracy: 0.7617 - lr: 9.2234e-06 - 1s/epoch - 4ms/step
Epoch 217/300
283/283 - 1s - loss: 0.5951 - accuracy: 0.7869 - val_loss: 0.8933 - val_accuracy: 0.7630 - lr: 9.2234e-06 - 1s/epoch - 4ms/step
Epoch 218/300
283/283 - 1s - loss: 0.5882 - accuracy: 0.7944 - val_loss: 0.8939 - val_accuracy: 0.7617 - lr: 9.2234e-06 - 1s/epoch - 4ms/step
Epoch 219/300
283/283 - 1s - loss: 0.5999 - accuracy: 0.7889 - val_loss: 0.9029 - val_accuracy: 0.7597 - lr: 9.2234e-06 - 1s/epoch - 4ms/step
Epoch 220/300
283/283 - 1s - loss: 0.5863 - accuracy: 0.7898 - val_loss: 0.9074 - val_accuracy: 0.7617 - lr: 9.2234e-06 - 1s/epoch - 4ms/step
Epoch 221/300
283/283 - 1s - loss: 0.5828 - accuracy: 0.7936 - val_loss: 0.9002 - val_accuracy: 0.7607 - lr: 7.3787e-06 - 1s/epoch - 4ms/step
Epoch 222/300
283/283 - 1s - loss: 0.5830 - accuracy: 0.7949 - val_loss: 0.8971 - val_accuracy: 0.7617 - lr: 7.3787e-06 - 1s/epoch - 4ms/step
Epoch 223/300
283/283 - 1s - loss: 0.5810 - accuracy: 0.7936 - val_loss: 0.9001 - val_accuracy: 0.7620 - lr: 7.3787e-06 - 1s/epoch - 4ms/step
Epoch 224/300
283/283 - 1s - loss: 0.5907 - accuracy: 0.7921 - val_loss: 0.9039 - val_accuracy: 0.7613 - lr: 7.3787e-06 - 1s/epoch - 4ms/step
Epoch 225/300
283/283 - 1s - loss: 0.5874 - accuracy: 0.7936 - val_loss: 0.8979 - val_accuracy: 0.7597 - lr: 7.3787e-06 - 1s/epoch - 4ms/step
Epoch 226/300
283/283 - 1s - loss: 0.5872 - accuracy: 0.7901 - val_loss: 0.8979 - val_accuracy: 0.7617 - lr: 7.3787e-06 - 1s/epoch - 4ms/step
Epoch 227/300
283/283 - 1s - loss: 0.5980 - accuracy: 0.7829 - val_loss: 0.8950 - val_accuracy: 0.7620 - lr: 7.3787e-06 - 1s/epoch - 4ms/step
Epoch 228/300
283/283 - 1s - loss: 0.5715 - accuracy: 0.7977 - val_loss: 0.9006 - val_accuracy: 0.7620 - lr: 7.3787e-06 - 1s/epoch - 4ms/step
Epoch 229/300
283/283 - 1s - loss: 0.5891 - accuracy: 0.7910 - val_loss: 0.8928 - val_accuracy: 0.7633 - lr: 7.3787e-06 - 1s/epoch - 4ms/step
Epoch 230/300
283/283 - 1s - loss: 0.5916 - accuracy: 0.7950 - val_loss: 0.9008 - val_accuracy: 0.7600 - lr: 7.3787e-06 - 1s/epoch - 4ms/step
Epoch 231/300
283/283 - 1s - loss: 0.5780 - accuracy: 0.7967 - val_loss: 0.8994 - val_accuracy: 0.7617 - lr: 5.9030e-06 - 1s/epoch - 4ms/step
Epoch 232/300
283/283 - 1s - loss: 0.5909 - accuracy: 0.7928 - val_loss: 0.8974 - val_accuracy: 0.7643 - lr: 5.9030e-06 - 1s/epoch - 4ms/step
Epoch 233/300
283/283 - 1s - loss: 0.6017 - accuracy: 0.7898 - val_loss: 0.8988 - val_accuracy: 0.7637 - lr: 5.9030e-06 - 1s/epoch - 4ms/step
Epoch 234/300
283/283 - 1s - loss: 0.5803 - accuracy: 0.7967 - val_loss: 0.9003 - val_accuracy: 0.7633 - lr: 5.9030e-06 - 1s/epoch - 4ms/step
Epoch 235/300
283/283 - 1s - loss: 0.5879 - accuracy: 0.7898 - val_loss: 0.8995 - val_accuracy: 0.7597 - lr: 5.9030e-06 - 1s/epoch - 4ms/step
Epoch 236/300
283/283 - 1s - loss: 0.5835 - accuracy: 0.7971 - val_loss: 0.9029 - val_accuracy: 0.7607 - lr: 5.9030e-06 - 1s/epoch - 4ms/step
Epoch 237/300
283/283 - 1s - loss: 0.5774 - accuracy: 0.7952 - val_loss: 0.9017 - val_accuracy: 0.7610 - lr: 5.9030e-06 - 1s/epoch - 4ms/step
Epoch 238/300
283/283 - 1s - loss: 0.5820 - accuracy: 0.7933 - val_loss: 0.9026 - val_accuracy: 0.7610 - lr: 5.9030e-06 - 1s/epoch - 4ms/step
Epoch 239/300
283/283 - 1s - loss: 0.5846 - accuracy: 0.7936 - val_loss: 0.9026 - val_accuracy: 0.7603 - lr: 5.9030e-06 - 1s/epoch - 4ms/step
Epoch 240/300
283/283 - 1s - loss: 0.5881 - accuracy: 0.7914 - val_loss: 0.9054 - val_accuracy: 0.7607 - lr: 5.9030e-06 - 1s/epoch - 4ms/step
Epoch 241/300
283/283 - 1s - loss: 0.5896 - accuracy: 0.7934 - val_loss: 0.9032 - val_accuracy: 0.7607 - lr: 4.7224e-06 - 1s/epoch - 4ms/step
Epoch 242/300
283/283 - 1s - loss: 0.5846 - accuracy: 0.7967 - val_loss: 0.9041 - val_accuracy: 0.7597 - lr: 4.7224e-06 - 1s/epoch - 4ms/step
Epoch 243/300
283/283 - 1s - loss: 0.5806 - accuracy: 0.7924 - val_loss: 0.8969 - val_accuracy: 0.7610 - lr: 4.7224e-06 - 1s/epoch - 4ms/step
Epoch 244/300
283/283 - 1s - loss: 0.5915 - accuracy: 0.7977 - val_loss: 0.9037 - val_accuracy: 0.7613 - lr: 4.7224e-06 - 1s/epoch - 4ms/step
Epoch 245/300
283/283 - 1s - loss: 0.5944 - accuracy: 0.7887 - val_loss: 0.8997 - val_accuracy: 0.7630 - lr: 4.7224e-06 - 1s/epoch - 4ms/step
Epoch 246/300
283/283 - 1s - loss: 0.5804 - accuracy: 0.7945 - val_loss: 0.8977 - val_accuracy: 0.7623 - lr: 4.7224e-06 - 1s/epoch - 4ms/step
Epoch 247/300
283/283 - 1s - loss: 0.5939 - accuracy: 0.7890 - val_loss: 0.9009 - val_accuracy: 0.7610 - lr: 4.7224e-06 - 1s/epoch - 4ms/step
Epoch 248/300
283/283 - 1s - loss: 0.5743 - accuracy: 0.7915 - val_loss: 0.8989 - val_accuracy: 0.7617 - lr: 4.7224e-06 - 1s/epoch - 4ms/step
Epoch 249/300
283/283 - 1s - loss: 0.5817 - accuracy: 0.7954 - val_loss: 0.9009 - val_accuracy: 0.7613 - lr: 4.7224e-06 - 1s/epoch - 4ms/step
Epoch 250/300
283/283 - 1s - loss: 0.5881 - accuracy: 0.7940 - val_loss: 0.8992 - val_accuracy: 0.7633 - lr: 4.7224e-06 - 1s/epoch - 4ms/step
Epoch 251/300
283/283 - 1s - loss: 0.5909 - accuracy: 0.7908 - val_loss: 0.9010 - val_accuracy: 0.7620 - lr: 3.7779e-06 - 1s/epoch - 4ms/step
Epoch 252/300
283/283 - 1s - loss: 0.5788 - accuracy: 0.7935 - val_loss: 0.9050 - val_accuracy: 0.7623 - lr: 3.7779e-06 - 1s/epoch - 4ms/step
Epoch 253/300
283/283 - 1s - loss: 0.5866 - accuracy: 0.7935 - val_loss: 0.8995 - val_accuracy: 0.7627 - lr: 3.7779e-06 - 1s/epoch - 4ms/step
Epoch 254/300
283/283 - 1s - loss: 0.5957 - accuracy: 0.7919 - val_loss: 0.9016 - val_accuracy: 0.7627 - lr: 3.7779e-06 - 1s/epoch - 4ms/step
Epoch 255/300
283/283 - 1s - loss: 0.5767 - accuracy: 0.7928 - val_loss: 0.9020 - val_accuracy: 0.7630 - lr: 3.7779e-06 - 1s/epoch - 4ms/step
Epoch 256/300
283/283 - 1s - loss: 0.5848 - accuracy: 0.7956 - val_loss: 0.9040 - val_accuracy: 0.7620 - lr: 3.7779e-06 - 1s/epoch - 4ms/step
Epoch 257/300
283/283 - 1s - loss: 0.5755 - accuracy: 0.7994 - val_loss: 0.9000 - val_accuracy: 0.7620 - lr: 3.7779e-06 - 1s/epoch - 4ms/step
Epoch 258/300
283/283 - 1s - loss: 0.5775 - accuracy: 0.7926 - val_loss: 0.9021 - val_accuracy: 0.7623 - lr: 3.7779e-06 - 1s/epoch - 4ms/step
Epoch 259/300
283/283 - 1s - loss: 0.5871 - accuracy: 0.7919 - val_loss: 0.9019 - val_accuracy: 0.7640 - lr: 3.7779e-06 - 1s/epoch - 4ms/step
Epoch 260/300
283/283 - 1s - loss: 0.5752 - accuracy: 0.7966 - val_loss: 0.9050 - val_accuracy: 0.7613 - lr: 3.7779e-06 - 1s/epoch - 4ms/step
Epoch 261/300
283/283 - 1s - loss: 0.5740 - accuracy: 0.7986 - val_loss: 0.8996 - val_accuracy: 0.7613 - lr: 3.0223e-06 - 1s/epoch - 4ms/step
Epoch 262/300
283/283 - 1s - loss: 0.5828 - accuracy: 0.7974 - val_loss: 0.8994 - val_accuracy: 0.7633 - lr: 3.0223e-06 - 1s/epoch - 4ms/step
Epoch 263/300
283/283 - 1s - loss: 0.5895 - accuracy: 0.7929 - val_loss: 0.9033 - val_accuracy: 0.7620 - lr: 3.0223e-06 - 1s/epoch - 4ms/step
Epoch 264/300
283/283 - 1s - loss: 0.5752 - accuracy: 0.7961 - val_loss: 0.9018 - val_accuracy: 0.7627 - lr: 3.0223e-06 - 1s/epoch - 4ms/step
Epoch 265/300
283/283 - 1s - loss: 0.5768 - accuracy: 0.7949 - val_loss: 0.9026 - val_accuracy: 0.7630 - lr: 3.0223e-06 - 1s/epoch - 4ms/step
Epoch 266/300
283/283 - 1s - loss: 0.5806 - accuracy: 0.7976 - val_loss: 0.9029 - val_accuracy: 0.7617 - lr: 3.0223e-06 - 1s/epoch - 4ms/step
Epoch 267/300
283/283 - 1s - loss: 0.5797 - accuracy: 0.7972 - val_loss: 0.9001 - val_accuracy: 0.7617 - lr: 3.0223e-06 - 1s/epoch - 4ms/step
Epoch 268/300
283/283 - 1s - loss: 0.5681 - accuracy: 0.8018 - val_loss: 0.9006 - val_accuracy: 0.7623 - lr: 3.0223e-06 - 1s/epoch - 4ms/step
Epoch 269/300
283/283 - 1s - loss: 0.5902 - accuracy: 0.7908 - val_loss: 0.9027 - val_accuracy: 0.7620 - lr: 3.0223e-06 - 1s/epoch - 4ms/step
Epoch 270/300
283/283 - 1s - loss: 0.5809 - accuracy: 0.7922 - val_loss: 0.8954 - val_accuracy: 0.7630 - lr: 3.0223e-06 - 1s/epoch - 4ms/step
Epoch 271/300
283/283 - 1s - loss: 0.5803 - accuracy: 0.7990 - val_loss: 0.9003 - val_accuracy: 0.7623 - lr: 2.4179e-06 - 1s/epoch - 4ms/step
Epoch 272/300
283/283 - 1s - loss: 0.5753 - accuracy: 0.7945 - val_loss: 0.9007 - val_accuracy: 0.7620 - lr: 2.4179e-06 - 1s/epoch - 4ms/step
Epoch 273/300
283/283 - 1s - loss: 0.5783 - accuracy: 0.7956 - val_loss: 0.9005 - val_accuracy: 0.7620 - lr: 2.4179e-06 - 1s/epoch - 4ms/step
Epoch 274/300
283/283 - 1s - loss: 0.5850 - accuracy: 0.7933 - val_loss: 0.9019 - val_accuracy: 0.7623 - lr: 2.4179e-06 - 1s/epoch - 4ms/step
Epoch 275/300
283/283 - 1s - loss: 0.5826 - accuracy: 0.7975 - val_loss: 0.9043 - val_accuracy: 0.7630 - lr: 2.4179e-06 - 1s/epoch - 4ms/step
Epoch 276/300
283/283 - 1s - loss: 0.5735 - accuracy: 0.7961 - val_loss: 0.9022 - val_accuracy: 0.7620 - lr: 2.4179e-06 - 1s/epoch - 4ms/step
Epoch 277/300
283/283 - 1s - loss: 0.5933 - accuracy: 0.7901 - val_loss: 0.8997 - val_accuracy: 0.7647 - lr: 2.4179e-06 - 1s/epoch - 5ms/step
Epoch 278/300
283/283 - 1s - loss: 0.5809 - accuracy: 0.7967 - val_loss: 0.9004 - val_accuracy: 0.7633 - lr: 2.4179e-06 - 1s/epoch - 4ms/step
Epoch 279/300
283/283 - 1s - loss: 0.5819 - accuracy: 0.7963 - val_loss: 0.9042 - val_accuracy: 0.7630 - lr: 2.4179e-06 - 1s/epoch - 4ms/step
Epoch 280/300
283/283 - 1s - loss: 0.5782 - accuracy: 0.7946 - val_loss: 0.8999 - val_accuracy: 0.7627 - lr: 2.4179e-06 - 1s/epoch - 4ms/step
Epoch 281/300
283/283 - 1s - loss: 0.5883 - accuracy: 0.7938 - val_loss: 0.9009 - val_accuracy: 0.7627 - lr: 1.9343e-06 - 1s/epoch - 4ms/step
Epoch 282/300
283/283 - 1s - loss: 0.5756 - accuracy: 0.7959 - val_loss: 0.9011 - val_accuracy: 0.7627 - lr: 1.9343e-06 - 1s/epoch - 4ms/step
Epoch 283/300
283/283 - 1s - loss: 0.5729 - accuracy: 0.7991 - val_loss: 0.9026 - val_accuracy: 0.7630 - lr: 1.9343e-06 - 1s/epoch - 4ms/step
Epoch 284/300
283/283 - 1s - loss: 0.5794 - accuracy: 0.7964 - val_loss: 0.9013 - val_accuracy: 0.7630 - lr: 1.9343e-06 - 1s/epoch - 4ms/step
Epoch 285/300
283/283 - 1s - loss: 0.5823 - accuracy: 0.7964 - val_loss: 0.9006 - val_accuracy: 0.7630 - lr: 1.9343e-06 - 1s/epoch - 4ms/step
Epoch 286/300
283/283 - 1s - loss: 0.5904 - accuracy: 0.7901 - val_loss: 0.8995 - val_accuracy: 0.7640 - lr: 1.9343e-06 - 1s/epoch - 4ms/step
Epoch 287/300
283/283 - 1s - loss: 0.5831 - accuracy: 0.7918 - val_loss: 0.9002 - val_accuracy: 0.7633 - lr: 1.9343e-06 - 1s/epoch - 4ms/step
Epoch 288/300
283/283 - 1s - loss: 0.5773 - accuracy: 0.7922 - val_loss: 0.9032 - val_accuracy: 0.7627 - lr: 1.9343e-06 - 1s/epoch - 4ms/step
Epoch 289/300
283/283 - 1s - loss: 0.5800 - accuracy: 0.7954 - val_loss: 0.9027 - val_accuracy: 0.7613 - lr: 1.9343e-06 - 1s/epoch - 4ms/step
Epoch 290/300
283/283 - 1s - loss: 0.5726 - accuracy: 0.7966 - val_loss: 0.9026 - val_accuracy: 0.7640 - lr: 1.9343e-06 - 1s/epoch - 4ms/step
Epoch 291/300
283/283 - 1s - loss: 0.5712 - accuracy: 0.7961 - val_loss: 0.9026 - val_accuracy: 0.7633 - lr: 1.5474e-06 - 1s/epoch - 4ms/step
Epoch 292/300
283/283 - 1s - loss: 0.5703 - accuracy: 0.8015 - val_loss: 0.9031 - val_accuracy: 0.7623 - lr: 1.5474e-06 - 1s/epoch - 4ms/step
Epoch 293/300
283/283 - 1s - loss: 0.5720 - accuracy: 0.7962 - val_loss: 0.9032 - val_accuracy: 0.7623 - lr: 1.5474e-06 - 1s/epoch - 4ms/step
Epoch 294/300
283/283 - 1s - loss: 0.5844 - accuracy: 0.7913 - val_loss: 0.9002 - val_accuracy: 0.7627 - lr: 1.5474e-06 - 1s/epoch - 4ms/step
Epoch 295/300
283/283 - 1s - loss: 0.5824 - accuracy: 0.7970 - val_loss: 0.9037 - val_accuracy: 0.7613 - lr: 1.5474e-06 - 1s/epoch - 4ms/step
Epoch 296/300
283/283 - 1s - loss: 0.5799 - accuracy: 0.7955 - val_loss: 0.9008 - val_accuracy: 0.7623 - lr: 1.5474e-06 - 1s/epoch - 4ms/step
Epoch 297/300
283/283 - 1s - loss: 0.5924 - accuracy: 0.7935 - val_loss: 0.9026 - val_accuracy: 0.7623 - lr: 1.5474e-06 - 1s/epoch - 4ms/step
Epoch 298/300
283/283 - 1s - loss: 0.5879 - accuracy: 0.7928 - val_loss: 0.9008 - val_accuracy: 0.7620 - lr: 1.5474e-06 - 1s/epoch - 4ms/step
Epoch 299/300
283/283 - 1s - loss: 0.5811 - accuracy: 0.7971 - val_loss: 0.8991 - val_accuracy: 0.7617 - lr: 1.5474e-06 - 1s/epoch - 4ms/step
Epoch 300/300
283/283 - 1s - loss: 0.5730 - accuracy: 0.7976 - val_loss: 0.9021 - val_accuracy: 0.7633 - lr: 1.5474e-06 - 1s/epoch - 4ms/step
94/94 - 0s - loss: 0.9021 - accuracy: 0.7633 - 197ms/epoch - 2ms/step
Baseline Error: 23.67%

We can see that the learning rate scheduler has not improved the performance of the model, the accuracy is stll flattening out after about 100 epochs and the overall accuracy has decreased as well. Instead of using the learning rate scheduler, we can try to add more dropout layers/regularization layers to RegularizedDOModel2 (300 epoch model) and see if we can overcome the overfitting problem and get a val accuracy of over 90%.

#Increase dropout

model_31new = models.Sequential([
    layers.Conv2D(32, (3, 3), activation='relu', input_shape=(31, 31, 1)),
    layers.MaxPooling2D((2, 2)),
    layers.Dropout(0.2),
    
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.MaxPooling2D((2, 2)),
    layers.Dropout(0.2),
    
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.MaxPooling2D((2, 2)),
    layers.Dropout(0.2),
    
    layers.Flatten(),
    
    layers.Dense(64, activation='relu', kernel_regularizer=regularizers.l2(0.001)),
    layers.Dropout(0.5),
    
    layers.Dense(15, activation='softmax')
])



model_31new.compile(optimizer='adam',
                loss= 'categorical_crossentropy',
                metrics=['accuracy'])

history = model_31new.fit(X_train31, train_labels, validation_data=(X_val31, validation_labels),
                    epochs=350, batch_size=200, verbose=2, class_weight = class_weight) #adjust epochs to 300

scores = model_31new.evaluate(X_val31, validation_labels, verbose=2)
print("Baseline Error: %.2f%%" % (100-scores[1]*100))
plot_history(history)
df31.loc[len(df31)] = ['ExtraRegularReducedModel', history.history['accuracy'][-1], history.history['val_accuracy'][-1], history.history['loss'][-1], history.history['val_loss'][-1], history]

Epoch 1/350

2023-11-26 13:22:32.809832: E tensorflow/core/grappler/optimizers/meta_optimizer.cc:954] layout failed: INVALID_ARGUMENT: Size of values 0 does not match size of permutation 4 @ fanin shape insequential_10/dropout_3/dropout/SelectV2-2-TransposeNHWCToNCHW-LayoutOptimizer

46/46 - 3s - loss: 3.2468 - accuracy: 0.1008 - val_loss: 2.8435 - val_accuracy: 0.0667 - 3s/epoch - 55ms/step
Epoch 2/350
46/46 - 0s - loss: 3.1411 - accuracy: 0.1131 - val_loss: 2.8132 - val_accuracy: 0.0890 - 247ms/epoch - 5ms/step
Epoch 3/350
46/46 - 0s - loss: 2.9888 - accuracy: 0.1508 - val_loss: 2.7089 - val_accuracy: 0.1040 - 243ms/epoch - 5ms/step
Epoch 4/350
46/46 - 0s - loss: 2.7869 - accuracy: 0.1918 - val_loss: 2.4833 - val_accuracy: 0.1900 - 240ms/epoch - 5ms/step
Epoch 5/350
46/46 - 0s - loss: 2.6165 - accuracy: 0.2307 - val_loss: 2.3508 - val_accuracy: 0.2273 - 245ms/epoch - 5ms/step
Epoch 6/350
46/46 - 0s - loss: 2.4703 - accuracy: 0.2661 - val_loss: 2.2466 - val_accuracy: 0.2527 - 244ms/epoch - 5ms/step
Epoch 7/350
46/46 - 0s - loss: 2.3777 - accuracy: 0.2869 - val_loss: 2.2395 - val_accuracy: 0.2827 - 244ms/epoch - 5ms/step
Epoch 8/350
46/46 - 0s - loss: 2.2827 - accuracy: 0.3101 - val_loss: 2.0958 - val_accuracy: 0.3100 - 240ms/epoch - 5ms/step
Epoch 9/350
46/46 - 0s - loss: 2.2064 - accuracy: 0.3290 - val_loss: 1.9944 - val_accuracy: 0.3603 - 252ms/epoch - 5ms/step
Epoch 10/350
46/46 - 0s - loss: 2.1284 - accuracy: 0.3520 - val_loss: 2.1160 - val_accuracy: 0.3227 - 242ms/epoch - 5ms/step
Epoch 11/350
46/46 - 0s - loss: 2.1018 - accuracy: 0.3547 - val_loss: 1.9260 - val_accuracy: 0.3733 - 240ms/epoch - 5ms/step
Epoch 12/350
46/46 - 0s - loss: 1.9981 - accuracy: 0.3786 - val_loss: 1.7960 - val_accuracy: 0.4197 - 242ms/epoch - 5ms/step
Epoch 13/350
46/46 - 0s - loss: 1.9659 - accuracy: 0.3949 - val_loss: 1.7705 - val_accuracy: 0.4243 - 245ms/epoch - 5ms/step
Epoch 14/350
46/46 - 0s - loss: 1.9096 - accuracy: 0.4164 - val_loss: 1.7205 - val_accuracy: 0.4417 - 244ms/epoch - 5ms/step
Epoch 15/350
46/46 - 0s - loss: 1.8337 - accuracy: 0.4337 - val_loss: 1.7015 - val_accuracy: 0.4413 - 243ms/epoch - 5ms/step
Epoch 16/350
46/46 - 0s - loss: 1.8078 - accuracy: 0.4502 - val_loss: 1.7536 - val_accuracy: 0.4377 - 241ms/epoch - 5ms/step
Epoch 17/350
46/46 - 0s - loss: 1.8060 - accuracy: 0.4478 - val_loss: 1.5842 - val_accuracy: 0.4797 - 248ms/epoch - 5ms/step
Epoch 18/350
46/46 - 0s - loss: 1.7473 - accuracy: 0.4576 - val_loss: 1.6520 - val_accuracy: 0.4537 - 243ms/epoch - 5ms/step
Epoch 19/350
46/46 - 0s - loss: 1.7102 - accuracy: 0.4696 - val_loss: 1.7044 - val_accuracy: 0.4433 - 254ms/epoch - 6ms/step
Epoch 20/350
46/46 - 0s - loss: 1.7259 - accuracy: 0.4745 - val_loss: 1.5072 - val_accuracy: 0.5077 - 243ms/epoch - 5ms/step
Epoch 21/350
46/46 - 0s - loss: 1.6548 - accuracy: 0.4868 - val_loss: 1.4666 - val_accuracy: 0.5077 - 250ms/epoch - 5ms/step
Epoch 22/350
46/46 - 0s - loss: 1.6362 - accuracy: 0.4971 - val_loss: 1.4606 - val_accuracy: 0.5290 - 247ms/epoch - 5ms/step
Epoch 23/350
46/46 - 0s - loss: 1.6036 - accuracy: 0.5102 - val_loss: 1.4073 - val_accuracy: 0.5483 - 244ms/epoch - 5ms/step
Epoch 24/350
46/46 - 0s - loss: 1.5639 - accuracy: 0.5163 - val_loss: 1.4303 - val_accuracy: 0.5487 - 242ms/epoch - 5ms/step
Epoch 25/350
46/46 - 0s - loss: 1.5452 - accuracy: 0.5270 - val_loss: 1.4346 - val_accuracy: 0.5543 - 250ms/epoch - 5ms/step
Epoch 26/350
46/46 - 0s - loss: 1.5323 - accuracy: 0.5235 - val_loss: 1.3705 - val_accuracy: 0.5610 - 244ms/epoch - 5ms/step
Epoch 27/350
46/46 - 0s - loss: 1.5323 - accuracy: 0.5255 - val_loss: 1.3357 - val_accuracy: 0.5830 - 240ms/epoch - 5ms/step
Epoch 28/350
46/46 - 0s - loss: 1.4803 - accuracy: 0.5440 - val_loss: 1.4471 - val_accuracy: 0.5243 - 242ms/epoch - 5ms/step
Epoch 29/350
46/46 - 0s - loss: 1.4817 - accuracy: 0.5470 - val_loss: 1.3039 - val_accuracy: 0.5863 - 244ms/epoch - 5ms/step
Epoch 30/350
46/46 - 0s - loss: 1.4616 - accuracy: 0.5524 - val_loss: 1.2480 - val_accuracy: 0.6200 - 242ms/epoch - 5ms/step
Epoch 31/350
46/46 - 0s - loss: 1.4450 - accuracy: 0.5529 - val_loss: 1.2977 - val_accuracy: 0.5923 - 241ms/epoch - 5ms/step
Epoch 32/350
46/46 - 0s - loss: 1.4147 - accuracy: 0.5598 - val_loss: 1.2770 - val_accuracy: 0.5957 - 240ms/epoch - 5ms/step
Epoch 33/350
46/46 - 0s - loss: 1.3882 - accuracy: 0.5702 - val_loss: 1.2337 - val_accuracy: 0.6157 - 244ms/epoch - 5ms/step
Epoch 34/350
46/46 - 0s - loss: 1.3790 - accuracy: 0.5741 - val_loss: 1.1660 - val_accuracy: 0.6413 - 242ms/epoch - 5ms/step
Epoch 35/350
46/46 - 0s - loss: 1.3643 - accuracy: 0.5774 - val_loss: 1.1805 - val_accuracy: 0.6360 - 241ms/epoch - 5ms/step
Epoch 36/350
46/46 - 0s - loss: 1.3580 - accuracy: 0.5851 - val_loss: 1.2278 - val_accuracy: 0.6303 - 240ms/epoch - 5ms/step
Epoch 37/350
46/46 - 0s - loss: 1.3631 - accuracy: 0.5852 - val_loss: 1.2596 - val_accuracy: 0.6023 - 242ms/epoch - 5ms/step
Epoch 38/350
46/46 - 0s - loss: 1.3502 - accuracy: 0.5853 - val_loss: 1.1525 - val_accuracy: 0.6503 - 247ms/epoch - 5ms/step
Epoch 39/350
46/46 - 0s - loss: 1.3249 - accuracy: 0.5897 - val_loss: 1.1284 - val_accuracy: 0.6477 - 240ms/epoch - 5ms/step
Epoch 40/350
46/46 - 0s - loss: 1.2970 - accuracy: 0.5929 - val_loss: 1.2462 - val_accuracy: 0.6093 - 240ms/epoch - 5ms/step
Epoch 41/350
46/46 - 0s - loss: 1.2930 - accuracy: 0.5998 - val_loss: 1.0883 - val_accuracy: 0.6717 - 246ms/epoch - 5ms/step
Epoch 42/350
46/46 - 0s - loss: 1.2526 - accuracy: 0.6185 - val_loss: 1.0909 - val_accuracy: 0.6720 - 245ms/epoch - 5ms/step
Epoch 43/350
46/46 - 0s - loss: 1.2776 - accuracy: 0.6134 - val_loss: 1.0798 - val_accuracy: 0.6693 - 242ms/epoch - 5ms/step
Epoch 44/350
46/46 - 0s - loss: 1.2413 - accuracy: 0.6220 - val_loss: 1.0314 - val_accuracy: 0.6923 - 242ms/epoch - 5ms/step
Epoch 45/350
46/46 - 0s - loss: 1.2589 - accuracy: 0.6120 - val_loss: 1.0632 - val_accuracy: 0.6790 - 245ms/epoch - 5ms/step
Epoch 46/350
46/46 - 0s - loss: 1.2244 - accuracy: 0.6235 - val_loss: 1.0246 - val_accuracy: 0.6950 - 242ms/epoch - 5ms/step
Epoch 47/350
46/46 - 0s - loss: 1.2171 - accuracy: 0.6239 - val_loss: 1.0467 - val_accuracy: 0.6767 - 240ms/epoch - 5ms/step
Epoch 48/350
46/46 - 0s - loss: 1.2088 - accuracy: 0.6279 - val_loss: 1.0200 - val_accuracy: 0.6987 - 242ms/epoch - 5ms/step
Epoch 49/350
46/46 - 0s - loss: 1.2179 - accuracy: 0.6210 - val_loss: 1.0128 - val_accuracy: 0.6997 - 250ms/epoch - 5ms/step
Epoch 50/350
46/46 - 0s - loss: 1.1944 - accuracy: 0.6286 - val_loss: 1.0022 - val_accuracy: 0.7070 - 255ms/epoch - 6ms/step
Epoch 51/350
46/46 - 0s - loss: 1.1802 - accuracy: 0.6378 - val_loss: 1.0823 - val_accuracy: 0.6697 - 257ms/epoch - 6ms/step
Epoch 52/350
46/46 - 0s - loss: 1.1843 - accuracy: 0.6390 - val_loss: 1.0104 - val_accuracy: 0.6963 - 254ms/epoch - 6ms/step
Epoch 53/350
46/46 - 0s - loss: 1.1677 - accuracy: 0.6477 - val_loss: 0.9967 - val_accuracy: 0.6947 - 246ms/epoch - 5ms/step
Epoch 54/350
46/46 - 0s - loss: 1.1670 - accuracy: 0.6423 - val_loss: 0.9654 - val_accuracy: 0.7090 - 246ms/epoch - 5ms/step
Epoch 55/350
46/46 - 0s - loss: 1.1384 - accuracy: 0.6495 - val_loss: 0.9239 - val_accuracy: 0.7330 - 242ms/epoch - 5ms/step
Epoch 56/350
46/46 - 0s - loss: 1.1349 - accuracy: 0.6486 - val_loss: 0.9304 - val_accuracy: 0.7227 - 243ms/epoch - 5ms/step
Epoch 57/350
46/46 - 0s - loss: 1.1033 - accuracy: 0.6639 - val_loss: 0.9530 - val_accuracy: 0.7187 - 242ms/epoch - 5ms/step
Epoch 58/350
46/46 - 0s - loss: 1.1264 - accuracy: 0.6594 - val_loss: 0.9445 - val_accuracy: 0.7253 - 246ms/epoch - 5ms/step
Epoch 59/350
46/46 - 0s - loss: 1.1297 - accuracy: 0.6560 - val_loss: 0.9802 - val_accuracy: 0.7157 - 242ms/epoch - 5ms/step
Epoch 60/350
46/46 - 0s - loss: 1.1413 - accuracy: 0.6472 - val_loss: 0.9564 - val_accuracy: 0.7200 - 241ms/epoch - 5ms/step
Epoch 61/350
46/46 - 0s - loss: 1.0997 - accuracy: 0.6612 - val_loss: 0.8935 - val_accuracy: 0.7397 - 248ms/epoch - 5ms/step
Epoch 62/350
46/46 - 0s - loss: 1.0727 - accuracy: 0.6722 - val_loss: 0.8909 - val_accuracy: 0.7387 - 246ms/epoch - 5ms/step
Epoch 63/350
46/46 - 0s - loss: 1.0796 - accuracy: 0.6659 - val_loss: 0.8840 - val_accuracy: 0.7433 - 244ms/epoch - 5ms/step
Epoch 64/350
46/46 - 0s - loss: 1.0903 - accuracy: 0.6700 - val_loss: 1.2769 - val_accuracy: 0.5977 - 244ms/epoch - 5ms/step
Epoch 65/350
46/46 - 0s - loss: 1.1252 - accuracy: 0.6534 - val_loss: 0.8908 - val_accuracy: 0.7360 - 244ms/epoch - 5ms/step
Epoch 66/350
46/46 - 0s - loss: 1.0650 - accuracy: 0.6743 - val_loss: 0.8781 - val_accuracy: 0.7423 - 256ms/epoch - 6ms/step
Epoch 67/350
46/46 - 0s - loss: 1.0783 - accuracy: 0.6694 - val_loss: 0.8542 - val_accuracy: 0.7553 - 254ms/epoch - 6ms/step
Epoch 68/350
46/46 - 0s - loss: 1.0270 - accuracy: 0.6860 - val_loss: 0.8860 - val_accuracy: 0.7340 - 256ms/epoch - 6ms/step
Epoch 69/350
46/46 - 0s - loss: 1.0432 - accuracy: 0.6870 - val_loss: 0.8968 - val_accuracy: 0.7367 - 253ms/epoch - 5ms/step
Epoch 70/350
46/46 - 0s - loss: 1.0531 - accuracy: 0.6810 - val_loss: 0.8475 - val_accuracy: 0.7580 - 245ms/epoch - 5ms/step
Epoch 71/350
46/46 - 0s - loss: 1.0094 - accuracy: 0.6901 - val_loss: 0.8410 - val_accuracy: 0.7603 - 246ms/epoch - 5ms/step
Epoch 72/350
46/46 - 0s - loss: 1.0229 - accuracy: 0.6938 - val_loss: 0.8442 - val_accuracy: 0.7523 - 243ms/epoch - 5ms/step
Epoch 73/350
46/46 - 0s - loss: 0.9935 - accuracy: 0.6938 - val_loss: 0.8663 - val_accuracy: 0.7450 - 245ms/epoch - 5ms/step
Epoch 74/350
46/46 - 0s - loss: 1.0148 - accuracy: 0.6932 - val_loss: 0.8460 - val_accuracy: 0.7557 - 245ms/epoch - 5ms/step
Epoch 75/350
46/46 - 0s - loss: 0.9912 - accuracy: 0.6992 - val_loss: 0.8248 - val_accuracy: 0.7647 - 242ms/epoch - 5ms/step
Epoch 76/350
46/46 - 0s - loss: 0.9855 - accuracy: 0.6930 - val_loss: 0.8420 - val_accuracy: 0.7543 - 239ms/epoch - 5ms/step
Epoch 77/350
46/46 - 0s - loss: 1.0055 - accuracy: 0.6977 - val_loss: 0.8307 - val_accuracy: 0.7557 - 242ms/epoch - 5ms/step
Epoch 78/350
46/46 - 0s - loss: 0.9725 - accuracy: 0.7047 - val_loss: 0.8236 - val_accuracy: 0.7597 - 246ms/epoch - 5ms/step
Epoch 79/350
46/46 - 0s - loss: 0.9769 - accuracy: 0.7018 - val_loss: 0.8398 - val_accuracy: 0.7490 - 243ms/epoch - 5ms/step
Epoch 80/350
46/46 - 0s - loss: 0.9723 - accuracy: 0.7072 - val_loss: 0.7925 - val_accuracy: 0.7697 - 240ms/epoch - 5ms/step
Epoch 81/350
46/46 - 0s - loss: 0.9680 - accuracy: 0.7005 - val_loss: 0.8133 - val_accuracy: 0.7620 - 247ms/epoch - 5ms/step
Epoch 82/350
46/46 - 0s - loss: 0.9512 - accuracy: 0.7074 - val_loss: 0.7694 - val_accuracy: 0.7763 - 243ms/epoch - 5ms/step
Epoch 83/350
46/46 - 0s - loss: 0.9481 - accuracy: 0.7051 - val_loss: 0.8036 - val_accuracy: 0.7607 - 245ms/epoch - 5ms/step
Epoch 84/350
46/46 - 0s - loss: 0.9287 - accuracy: 0.7142 - val_loss: 0.8937 - val_accuracy: 0.7297 - 242ms/epoch - 5ms/step
Epoch 85/350
46/46 - 0s - loss: 0.9708 - accuracy: 0.7016 - val_loss: 0.7947 - val_accuracy: 0.7670 - 241ms/epoch - 5ms/step
Epoch 86/350
46/46 - 0s - loss: 0.9316 - accuracy: 0.7147 - val_loss: 0.7725 - val_accuracy: 0.7777 - 241ms/epoch - 5ms/step
Epoch 87/350
46/46 - 0s - loss: 0.9433 - accuracy: 0.7188 - val_loss: 0.7656 - val_accuracy: 0.7777 - 246ms/epoch - 5ms/step
Epoch 88/350
46/46 - 0s - loss: 0.9214 - accuracy: 0.7229 - val_loss: 0.7519 - val_accuracy: 0.7830 - 246ms/epoch - 5ms/step
Epoch 89/350
46/46 - 0s - loss: 0.9417 - accuracy: 0.7141 - val_loss: 0.7866 - val_accuracy: 0.7680 - 248ms/epoch - 5ms/step
Epoch 90/350
46/46 - 0s - loss: 0.9354 - accuracy: 0.7107 - val_loss: 0.7624 - val_accuracy: 0.7777 - 249ms/epoch - 5ms/step
Epoch 91/350
46/46 - 0s - loss: 0.8961 - accuracy: 0.7302 - val_loss: 0.7378 - val_accuracy: 0.7887 - 244ms/epoch - 5ms/step
Epoch 92/350
46/46 - 0s - loss: 0.9062 - accuracy: 0.7232 - val_loss: 0.7733 - val_accuracy: 0.7743 - 242ms/epoch - 5ms/step
Epoch 93/350
46/46 - 0s - loss: 0.8958 - accuracy: 0.7257 - val_loss: 0.7510 - val_accuracy: 0.7820 - 242ms/epoch - 5ms/step
Epoch 94/350
46/46 - 0s - loss: 0.9746 - accuracy: 0.7057 - val_loss: 0.7505 - val_accuracy: 0.7927 - 245ms/epoch - 5ms/step
Epoch 95/350
46/46 - 0s - loss: 0.8852 - accuracy: 0.7244 - val_loss: 0.7287 - val_accuracy: 0.7897 - 246ms/epoch - 5ms/step
Epoch 96/350
46/46 - 0s - loss: 0.8861 - accuracy: 0.7252 - val_loss: 0.7203 - val_accuracy: 0.7940 - 242ms/epoch - 5ms/step
Epoch 97/350
46/46 - 0s - loss: 0.8844 - accuracy: 0.7265 - val_loss: 0.7215 - val_accuracy: 0.7940 - 248ms/epoch - 5ms/step
Epoch 98/350
46/46 - 0s - loss: 0.8706 - accuracy: 0.7304 - val_loss: 0.7400 - val_accuracy: 0.7887 - 245ms/epoch - 5ms/step
Epoch 99/350
46/46 - 0s - loss: 0.8862 - accuracy: 0.7286 - val_loss: 0.7475 - val_accuracy: 0.7893 - 243ms/epoch - 5ms/step
Epoch 100/350
46/46 - 0s - loss: 0.8743 - accuracy: 0.7346 - val_loss: 0.7994 - val_accuracy: 0.7660 - 242ms/epoch - 5ms/step
Epoch 101/350
46/46 - 0s - loss: 0.8684 - accuracy: 0.7322 - val_loss: 0.7235 - val_accuracy: 0.7967 - 245ms/epoch - 5ms/step
Epoch 102/350
46/46 - 0s - loss: 0.8674 - accuracy: 0.7368 - val_loss: 0.6968 - val_accuracy: 0.8063 - 243ms/epoch - 5ms/step
Epoch 103/350
46/46 - 0s - loss: 0.8720 - accuracy: 0.7321 - val_loss: 0.7538 - val_accuracy: 0.7893 - 239ms/epoch - 5ms/step
Epoch 104/350
46/46 - 0s - loss: 0.8629 - accuracy: 0.7376 - val_loss: 0.7065 - val_accuracy: 0.8043 - 239ms/epoch - 5ms/step
Epoch 105/350
46/46 - 0s - loss: 0.8457 - accuracy: 0.7420 - val_loss: 0.7242 - val_accuracy: 0.7947 - 241ms/epoch - 5ms/step
Epoch 106/350
46/46 - 0s - loss: 0.8803 - accuracy: 0.7333 - val_loss: 0.7148 - val_accuracy: 0.7973 - 242ms/epoch - 5ms/step
Epoch 107/350
46/46 - 0s - loss: 0.8432 - accuracy: 0.7462 - val_loss: 0.7006 - val_accuracy: 0.7987 - 242ms/epoch - 5ms/step
Epoch 108/350
46/46 - 0s - loss: 0.8544 - accuracy: 0.7409 - val_loss: 0.6917 - val_accuracy: 0.7950 - 240ms/epoch - 5ms/step
Epoch 109/350
46/46 - 0s - loss: 0.8326 - accuracy: 0.7418 - val_loss: 0.6869 - val_accuracy: 0.8037 - 243ms/epoch - 5ms/step
Epoch 110/350
46/46 - 0s - loss: 0.8433 - accuracy: 0.7424 - val_loss: 0.6745 - val_accuracy: 0.8083 - 253ms/epoch - 5ms/step
Epoch 111/350
46/46 - 0s - loss: 0.8312 - accuracy: 0.7478 - val_loss: 0.6887 - val_accuracy: 0.8010 - 258ms/epoch - 6ms/step
Epoch 112/350
46/46 - 0s - loss: 0.8152 - accuracy: 0.7461 - val_loss: 0.6704 - val_accuracy: 0.8087 - 261ms/epoch - 6ms/step
Epoch 113/350
46/46 - 0s - loss: 0.8369 - accuracy: 0.7446 - val_loss: 0.7940 - val_accuracy: 0.7637 - 254ms/epoch - 6ms/step
Epoch 114/350
46/46 - 0s - loss: 0.8376 - accuracy: 0.7416 - val_loss: 0.6956 - val_accuracy: 0.8023 - 259ms/epoch - 6ms/step
Epoch 115/350
46/46 - 0s - loss: 0.8307 - accuracy: 0.7477 - val_loss: 0.6604 - val_accuracy: 0.8140 - 268ms/epoch - 6ms/step
Epoch 116/350
46/46 - 0s - loss: 0.8360 - accuracy: 0.7449 - val_loss: 0.6858 - val_accuracy: 0.8023 - 258ms/epoch - 6ms/step
Epoch 117/350
46/46 - 0s - loss: 0.7810 - accuracy: 0.7622 - val_loss: 0.6705 - val_accuracy: 0.8083 - 264ms/epoch - 6ms/step
Epoch 118/350
46/46 - 0s - loss: 0.7949 - accuracy: 0.7574 - val_loss: 0.6625 - val_accuracy: 0.8150 - 268ms/epoch - 6ms/step
Epoch 119/350
46/46 - 0s - loss: 0.8130 - accuracy: 0.7489 - val_loss: 0.7155 - val_accuracy: 0.7917 - 264ms/epoch - 6ms/step
Epoch 120/350
46/46 - 0s - loss: 0.8081 - accuracy: 0.7573 - val_loss: 0.6485 - val_accuracy: 0.8170 - 265ms/epoch - 6ms/step
Epoch 121/350
46/46 - 0s - loss: 0.7843 - accuracy: 0.7563 - val_loss: 0.6609 - val_accuracy: 0.8077 - 268ms/epoch - 6ms/step
Epoch 122/350
46/46 - 0s - loss: 0.7977 - accuracy: 0.7583 - val_loss: 0.6477 - val_accuracy: 0.8123 - 254ms/epoch - 6ms/step
Epoch 123/350
46/46 - 0s - loss: 0.7771 - accuracy: 0.7616 - val_loss: 0.6632 - val_accuracy: 0.8097 - 255ms/epoch - 6ms/step
Epoch 124/350
46/46 - 0s - loss: 0.7840 - accuracy: 0.7550 - val_loss: 0.6514 - val_accuracy: 0.8163 - 253ms/epoch - 6ms/step
Epoch 125/350
46/46 - 0s - loss: 0.7875 - accuracy: 0.7624 - val_loss: 0.7019 - val_accuracy: 0.7953 - 244ms/epoch - 5ms/step
Epoch 126/350
46/46 - 0s - loss: 0.8022 - accuracy: 0.7565 - val_loss: 0.6391 - val_accuracy: 0.8220 - 241ms/epoch - 5ms/step
Epoch 127/350
46/46 - 0s - loss: 0.7669 - accuracy: 0.7604 - val_loss: 0.6367 - val_accuracy: 0.8197 - 244ms/epoch - 5ms/step
Epoch 128/350
46/46 - 0s - loss: 0.7712 - accuracy: 0.7641 - val_loss: 0.7038 - val_accuracy: 0.8023 - 240ms/epoch - 5ms/step
Epoch 129/350
46/46 - 0s - loss: 0.8360 - accuracy: 0.7480 - val_loss: 0.6518 - val_accuracy: 0.8113 - 249ms/epoch - 5ms/step
Epoch 130/350
46/46 - 0s - loss: 0.7651 - accuracy: 0.7626 - val_loss: 0.6200 - val_accuracy: 0.8263 - 243ms/epoch - 5ms/step
Epoch 131/350
46/46 - 0s - loss: 0.7525 - accuracy: 0.7724 - val_loss: 0.6338 - val_accuracy: 0.8200 - 240ms/epoch - 5ms/step
Epoch 132/350
46/46 - 0s - loss: 0.7413 - accuracy: 0.7718 - val_loss: 0.6494 - val_accuracy: 0.8167 - 241ms/epoch - 5ms/step
Epoch 133/350
46/46 - 0s - loss: 0.7442 - accuracy: 0.7722 - val_loss: 0.6399 - val_accuracy: 0.8177 - 246ms/epoch - 5ms/step
Epoch 134/350
46/46 - 0s - loss: 0.7510 - accuracy: 0.7718 - val_loss: 0.6158 - val_accuracy: 0.8250 - 242ms/epoch - 5ms/step
Epoch 135/350
46/46 - 0s - loss: 0.7413 - accuracy: 0.7764 - val_loss: 0.6265 - val_accuracy: 0.8273 - 250ms/epoch - 5ms/step
Epoch 136/350
46/46 - 0s - loss: 0.7419 - accuracy: 0.7774 - val_loss: 0.5995 - val_accuracy: 0.8347 - 245ms/epoch - 5ms/step
Epoch 137/350
46/46 - 0s - loss: 0.7411 - accuracy: 0.7751 - val_loss: 0.6326 - val_accuracy: 0.8137 - 244ms/epoch - 5ms/step
Epoch 138/350
46/46 - 0s - loss: 0.7502 - accuracy: 0.7682 - val_loss: 0.6208 - val_accuracy: 0.8267 - 241ms/epoch - 5ms/step
Epoch 139/350
46/46 - 0s - loss: 0.7275 - accuracy: 0.7776 - val_loss: 0.6548 - val_accuracy: 0.8100 - 251ms/epoch - 5ms/step
Epoch 140/350
46/46 - 0s - loss: 0.7375 - accuracy: 0.7722 - val_loss: 0.6116 - val_accuracy: 0.8290 - 245ms/epoch - 5ms/step
Epoch 141/350
46/46 - 0s - loss: 0.7237 - accuracy: 0.7841 - val_loss: 0.6508 - val_accuracy: 0.8267 - 252ms/epoch - 5ms/step
Epoch 142/350
46/46 - 0s - loss: 0.7240 - accuracy: 0.7820 - val_loss: 0.6455 - val_accuracy: 0.8247 - 243ms/epoch - 5ms/step
Epoch 143/350
46/46 - 0s - loss: 0.7204 - accuracy: 0.7807 - val_loss: 0.6186 - val_accuracy: 0.8303 - 248ms/epoch - 5ms/step
Epoch 144/350
46/46 - 0s - loss: 0.7015 - accuracy: 0.7856 - val_loss: 0.6167 - val_accuracy: 0.8227 - 244ms/epoch - 5ms/step
Epoch 145/350
46/46 - 0s - loss: 0.7162 - accuracy: 0.7794 - val_loss: 0.5856 - val_accuracy: 0.8427 - 253ms/epoch - 6ms/step
Epoch 146/350
46/46 - 0s - loss: 0.6814 - accuracy: 0.7955 - val_loss: 0.5876 - val_accuracy: 0.8333 - 244ms/epoch - 5ms/step
Epoch 147/350
46/46 - 0s - loss: 0.6902 - accuracy: 0.7919 - val_loss: 0.5707 - val_accuracy: 0.8387 - 243ms/epoch - 5ms/step
Epoch 148/350
46/46 - 0s - loss: 0.6950 - accuracy: 0.7878 - val_loss: 0.5874 - val_accuracy: 0.8347 - 238ms/epoch - 5ms/step
Epoch 149/350
46/46 - 0s - loss: 0.7212 - accuracy: 0.7826 - val_loss: 0.6099 - val_accuracy: 0.8283 - 242ms/epoch - 5ms/step
Epoch 150/350
46/46 - 0s - loss: 0.6843 - accuracy: 0.7877 - val_loss: 0.5873 - val_accuracy: 0.8410 - 245ms/epoch - 5ms/step
Epoch 151/350
46/46 - 0s - loss: 0.6986 - accuracy: 0.7900 - val_loss: 0.5860 - val_accuracy: 0.8400 - 244ms/epoch - 5ms/step
Epoch 152/350
46/46 - 0s - loss: 0.6752 - accuracy: 0.7926 - val_loss: 0.5573 - val_accuracy: 0.8520 - 240ms/epoch - 5ms/step
Epoch 153/350
46/46 - 0s - loss: 0.7027 - accuracy: 0.7869 - val_loss: 0.6049 - val_accuracy: 0.8323 - 243ms/epoch - 5ms/step
Epoch 154/350
46/46 - 0s - loss: 0.6819 - accuracy: 0.7874 - val_loss: 0.6242 - val_accuracy: 0.8303 - 240ms/epoch - 5ms/step
Epoch 155/350
46/46 - 0s - loss: 0.6959 - accuracy: 0.7905 - val_loss: 0.6083 - val_accuracy: 0.8307 - 239ms/epoch - 5ms/step
Epoch 156/350
46/46 - 0s - loss: 0.6994 - accuracy: 0.7882 - val_loss: 0.5803 - val_accuracy: 0.8390 - 243ms/epoch - 5ms/step
Epoch 157/350
46/46 - 0s - loss: 0.7039 - accuracy: 0.7840 - val_loss: 0.5995 - val_accuracy: 0.8393 - 246ms/epoch - 5ms/step
Epoch 158/350
46/46 - 0s - loss: 0.6793 - accuracy: 0.7920 - val_loss: 0.5796 - val_accuracy: 0.8470 - 244ms/epoch - 5ms/step
Epoch 159/350
46/46 - 0s - loss: 0.6742 - accuracy: 0.7953 - val_loss: 0.5821 - val_accuracy: 0.8410 - 244ms/epoch - 5ms/step
Epoch 160/350
46/46 - 0s - loss: 0.6630 - accuracy: 0.7967 - val_loss: 0.6033 - val_accuracy: 0.8347 - 237ms/epoch - 5ms/step
Epoch 161/350
46/46 - 0s - loss: 0.6663 - accuracy: 0.7990 - val_loss: 0.5501 - val_accuracy: 0.8520 - 239ms/epoch - 5ms/step
Epoch 162/350
46/46 - 0s - loss: 0.6539 - accuracy: 0.8013 - val_loss: 0.5571 - val_accuracy: 0.8493 - 241ms/epoch - 5ms/step
Epoch 163/350
46/46 - 0s - loss: 0.6673 - accuracy: 0.7964 - val_loss: 0.5670 - val_accuracy: 0.8450 - 243ms/epoch - 5ms/step
Epoch 164/350
46/46 - 0s - loss: 0.6865 - accuracy: 0.7918 - val_loss: 0.5945 - val_accuracy: 0.8390 - 241ms/epoch - 5ms/step
Epoch 165/350
46/46 - 0s - loss: 0.6931 - accuracy: 0.7880 - val_loss: 0.6023 - val_accuracy: 0.8267 - 240ms/epoch - 5ms/step
Epoch 166/350
46/46 - 0s - loss: 0.6628 - accuracy: 0.7940 - val_loss: 0.5559 - val_accuracy: 0.8473 - 239ms/epoch - 5ms/step
Epoch 167/350
46/46 - 0s - loss: 0.6623 - accuracy: 0.8024 - val_loss: 0.5719 - val_accuracy: 0.8400 - 243ms/epoch - 5ms/step
Epoch 168/350
46/46 - 0s - loss: 0.6568 - accuracy: 0.7966 - val_loss: 0.6010 - val_accuracy: 0.8417 - 240ms/epoch - 5ms/step
Epoch 169/350
46/46 - 0s - loss: 0.6480 - accuracy: 0.8057 - val_loss: 0.5490 - val_accuracy: 0.8553 - 239ms/epoch - 5ms/step
Epoch 170/350
46/46 - 0s - loss: 0.6461 - accuracy: 0.8060 - val_loss: 0.5375 - val_accuracy: 0.8517 - 246ms/epoch - 5ms/step
Epoch 171/350
46/46 - 0s - loss: 0.6370 - accuracy: 0.8097 - val_loss: 0.5369 - val_accuracy: 0.8560 - 241ms/epoch - 5ms/step
Epoch 172/350
46/46 - 0s - loss: 0.6366 - accuracy: 0.8077 - val_loss: 0.5513 - val_accuracy: 0.8503 - 249ms/epoch - 5ms/step
Epoch 173/350
46/46 - 0s - loss: 0.6415 - accuracy: 0.8068 - val_loss: 0.5457 - val_accuracy: 0.8530 - 250ms/epoch - 5ms/step
Epoch 174/350
46/46 - 0s - loss: 0.6297 - accuracy: 0.8124 - val_loss: 0.5441 - val_accuracy: 0.8517 - 250ms/epoch - 5ms/step
Epoch 175/350
46/46 - 0s - loss: 0.6276 - accuracy: 0.8069 - val_loss: 0.5673 - val_accuracy: 0.8460 - 253ms/epoch - 6ms/step
Epoch 176/350
46/46 - 0s - loss: 0.6507 - accuracy: 0.8028 - val_loss: 0.5491 - val_accuracy: 0.8557 - 244ms/epoch - 5ms/step
Epoch 177/350
46/46 - 0s - loss: 0.6474 - accuracy: 0.8051 - val_loss: 0.5657 - val_accuracy: 0.8443 - 238ms/epoch - 5ms/step
Epoch 178/350
46/46 - 0s - loss: 0.6364 - accuracy: 0.8062 - val_loss: 0.5819 - val_accuracy: 0.8387 - 238ms/epoch - 5ms/step
Epoch 179/350
46/46 - 0s - loss: 0.6423 - accuracy: 0.8025 - val_loss: 0.5388 - val_accuracy: 0.8567 - 239ms/epoch - 5ms/step
Epoch 180/350
46/46 - 0s - loss: 0.6271 - accuracy: 0.8115 - val_loss: 0.5529 - val_accuracy: 0.8520 - 238ms/epoch - 5ms/step
Epoch 181/350
46/46 - 0s - loss: 0.6506 - accuracy: 0.7993 - val_loss: 0.5674 - val_accuracy: 0.8440 - 239ms/epoch - 5ms/step
Epoch 182/350
46/46 - 0s - loss: 0.6317 - accuracy: 0.8059 - val_loss: 0.5439 - val_accuracy: 0.8537 - 242ms/epoch - 5ms/step
Epoch 183/350
46/46 - 0s - loss: 0.6429 - accuracy: 0.8072 - val_loss: 0.5519 - val_accuracy: 0.8500 - 241ms/epoch - 5ms/step
Epoch 184/350
46/46 - 0s - loss: 0.6037 - accuracy: 0.8176 - val_loss: 0.5492 - val_accuracy: 0.8557 - 239ms/epoch - 5ms/step
Epoch 185/350
46/46 - 0s - loss: 0.6138 - accuracy: 0.8167 - val_loss: 0.5682 - val_accuracy: 0.8467 - 238ms/epoch - 5ms/step
Epoch 186/350
46/46 - 0s - loss: 0.6311 - accuracy: 0.8042 - val_loss: 0.5277 - val_accuracy: 0.8603 - 254ms/epoch - 6ms/step
Epoch 187/350
46/46 - 0s - loss: 0.6083 - accuracy: 0.8115 - val_loss: 0.5398 - val_accuracy: 0.8550 - 249ms/epoch - 5ms/step
Epoch 188/350
46/46 - 0s - loss: 0.6377 - accuracy: 0.8070 - val_loss: 0.5392 - val_accuracy: 0.8553 - 248ms/epoch - 5ms/step
Epoch 189/350
46/46 - 0s - loss: 0.6331 - accuracy: 0.8091 - val_loss: 0.5412 - val_accuracy: 0.8547 - 247ms/epoch - 5ms/step
Epoch 190/350
46/46 - 0s - loss: 0.5909 - accuracy: 0.8186 - val_loss: 0.5178 - val_accuracy: 0.8637 - 242ms/epoch - 5ms/step
Epoch 191/350
46/46 - 0s - loss: 0.6130 - accuracy: 0.8155 - val_loss: 0.5232 - val_accuracy: 0.8577 - 240ms/epoch - 5ms/step
Epoch 192/350
46/46 - 0s - loss: 0.6409 - accuracy: 0.8115 - val_loss: 0.6490 - val_accuracy: 0.8230 - 242ms/epoch - 5ms/step
Epoch 193/350
46/46 - 0s - loss: 0.6312 - accuracy: 0.8084 - val_loss: 0.5129 - val_accuracy: 0.8610 - 239ms/epoch - 5ms/step
Epoch 194/350
46/46 - 0s - loss: 0.5953 - accuracy: 0.8223 - val_loss: 0.5211 - val_accuracy: 0.8637 - 242ms/epoch - 5ms/step
Epoch 195/350
46/46 - 0s - loss: 0.5982 - accuracy: 0.8163 - val_loss: 0.5295 - val_accuracy: 0.8590 - 238ms/epoch - 5ms/step
Epoch 196/350
46/46 - 0s - loss: 0.5964 - accuracy: 0.8217 - val_loss: 0.5392 - val_accuracy: 0.8587 - 239ms/epoch - 5ms/step
Epoch 197/350
46/46 - 0s - loss: 0.6002 - accuracy: 0.8176 - val_loss: 0.5306 - val_accuracy: 0.8550 - 236ms/epoch - 5ms/step
Epoch 198/350
46/46 - 0s - loss: 0.6144 - accuracy: 0.8170 - val_loss: 0.5143 - val_accuracy: 0.8610 - 239ms/epoch - 5ms/step
Epoch 199/350
46/46 - 0s - loss: 0.5966 - accuracy: 0.8243 - val_loss: 0.5185 - val_accuracy: 0.8667 - 245ms/epoch - 5ms/step
Epoch 200/350
46/46 - 0s - loss: 0.5853 - accuracy: 0.8248 - val_loss: 0.5129 - val_accuracy: 0.8657 - 247ms/epoch - 5ms/step
Epoch 201/350
46/46 - 0s - loss: 0.5833 - accuracy: 0.8234 - val_loss: 0.5227 - val_accuracy: 0.8590 - 241ms/epoch - 5ms/step
Epoch 202/350
46/46 - 0s - loss: 0.6008 - accuracy: 0.8173 - val_loss: 0.5252 - val_accuracy: 0.8593 - 241ms/epoch - 5ms/step
Epoch 203/350
46/46 - 0s - loss: 0.6026 - accuracy: 0.8148 - val_loss: 0.5153 - val_accuracy: 0.8607 - 244ms/epoch - 5ms/step
Epoch 204/350
46/46 - 0s - loss: 0.5722 - accuracy: 0.8281 - val_loss: 0.5074 - val_accuracy: 0.8643 - 241ms/epoch - 5ms/step
Epoch 205/350
46/46 - 0s - loss: 0.5787 - accuracy: 0.8233 - val_loss: 0.5032 - val_accuracy: 0.8697 - 240ms/epoch - 5ms/step
Epoch 206/350
46/46 - 0s - loss: 0.5803 - accuracy: 0.8235 - val_loss: 0.5102 - val_accuracy: 0.8623 - 242ms/epoch - 5ms/step
Epoch 207/350
46/46 - 0s - loss: 0.6055 - accuracy: 0.8206 - val_loss: 0.5168 - val_accuracy: 0.8590 - 242ms/epoch - 5ms/step
Epoch 208/350
46/46 - 0s - loss: 0.5630 - accuracy: 0.8315 - val_loss: 0.5037 - val_accuracy: 0.8670 - 245ms/epoch - 5ms/step
Epoch 209/350
46/46 - 0s - loss: 0.5731 - accuracy: 0.8241 - val_loss: 0.4937 - val_accuracy: 0.8740 - 243ms/epoch - 5ms/step
Epoch 210/350
46/46 - 0s - loss: 0.5647 - accuracy: 0.8300 - val_loss: 0.5173 - val_accuracy: 0.8630 - 250ms/epoch - 5ms/step
Epoch 211/350
46/46 - 0s - loss: 0.5658 - accuracy: 0.8299 - val_loss: 0.5028 - val_accuracy: 0.8643 - 255ms/epoch - 6ms/step
Epoch 212/350
46/46 - 0s - loss: 0.5664 - accuracy: 0.8242 - val_loss: 0.5014 - val_accuracy: 0.8660 - 249ms/epoch - 5ms/step
Epoch 213/350
46/46 - 0s - loss: 0.5633 - accuracy: 0.8331 - val_loss: 0.5104 - val_accuracy: 0.8603 - 240ms/epoch - 5ms/step
Epoch 214/350
46/46 - 0s - loss: 0.5994 - accuracy: 0.8221 - val_loss: 0.5012 - val_accuracy: 0.8640 - 243ms/epoch - 5ms/step
Epoch 215/350
46/46 - 0s - loss: 0.5756 - accuracy: 0.8253 - val_loss: 0.5159 - val_accuracy: 0.8663 - 242ms/epoch - 5ms/step
Epoch 216/350
46/46 - 0s - loss: 0.5690 - accuracy: 0.8293 - val_loss: 0.5047 - val_accuracy: 0.8630 - 241ms/epoch - 5ms/step
Epoch 217/350
46/46 - 0s - loss: 0.5554 - accuracy: 0.8329 - val_loss: 0.4871 - val_accuracy: 0.8743 - 241ms/epoch - 5ms/step
Epoch 218/350
46/46 - 0s - loss: 0.5588 - accuracy: 0.8324 - val_loss: 0.4792 - val_accuracy: 0.8740 - 241ms/epoch - 5ms/step
Epoch 219/350
46/46 - 0s - loss: 0.5562 - accuracy: 0.8321 - val_loss: 0.4843 - val_accuracy: 0.8730 - 246ms/epoch - 5ms/step
Epoch 220/350
46/46 - 0s - loss: 0.5344 - accuracy: 0.8356 - val_loss: 0.5049 - val_accuracy: 0.8660 - 245ms/epoch - 5ms/step
Epoch 221/350
46/46 - 0s - loss: 0.5463 - accuracy: 0.8341 - val_loss: 0.4854 - val_accuracy: 0.8733 - 242ms/epoch - 5ms/step
Epoch 222/350
46/46 - 0s - loss: 0.6018 - accuracy: 0.8212 - val_loss: 0.5104 - val_accuracy: 0.8650 - 243ms/epoch - 5ms/step
Epoch 223/350
46/46 - 0s - loss: 0.5625 - accuracy: 0.8241 - val_loss: 0.4881 - val_accuracy: 0.8713 - 244ms/epoch - 5ms/step
Epoch 224/350
46/46 - 0s - loss: 0.5283 - accuracy: 0.8411 - val_loss: 0.4817 - val_accuracy: 0.8720 - 242ms/epoch - 5ms/step
Epoch 225/350
46/46 - 0s - loss: 0.5327 - accuracy: 0.8372 - val_loss: 0.4810 - val_accuracy: 0.8727 - 244ms/epoch - 5ms/step
Epoch 226/350
46/46 - 0s - loss: 0.5507 - accuracy: 0.8377 - val_loss: 0.4689 - val_accuracy: 0.8763 - 253ms/epoch - 5ms/step
Epoch 227/350
46/46 - 0s - loss: 0.5542 - accuracy: 0.8313 - val_loss: 0.4903 - val_accuracy: 0.8683 - 253ms/epoch - 6ms/step
Epoch 228/350
46/46 - 0s - loss: 0.5477 - accuracy: 0.8354 - val_loss: 0.4733 - val_accuracy: 0.8753 - 242ms/epoch - 5ms/step
Epoch 229/350
46/46 - 0s - loss: 0.5189 - accuracy: 0.8423 - val_loss: 0.4690 - val_accuracy: 0.8770 - 239ms/epoch - 5ms/step
Epoch 230/350
46/46 - 0s - loss: 0.5364 - accuracy: 0.8393 - val_loss: 0.4693 - val_accuracy: 0.8763 - 240ms/epoch - 5ms/step
Epoch 231/350
46/46 - 0s - loss: 0.5458 - accuracy: 0.8435 - val_loss: 0.4883 - val_accuracy: 0.8713 - 245ms/epoch - 5ms/step
Epoch 232/350
46/46 - 0s - loss: 0.5212 - accuracy: 0.8426 - val_loss: 0.4991 - val_accuracy: 0.8680 - 241ms/epoch - 5ms/step
Epoch 233/350
46/46 - 0s - loss: 0.5804 - accuracy: 0.8279 - val_loss: 0.4797 - val_accuracy: 0.8753 - 247ms/epoch - 5ms/step
Epoch 234/350
46/46 - 0s - loss: 0.5464 - accuracy: 0.8386 - val_loss: 0.4625 - val_accuracy: 0.8763 - 246ms/epoch - 5ms/step
Epoch 235/350
46/46 - 0s - loss: 0.5204 - accuracy: 0.8467 - val_loss: 0.4846 - val_accuracy: 0.8700 - 247ms/epoch - 5ms/step
Epoch 236/350
46/46 - 0s - loss: 0.5337 - accuracy: 0.8370 - val_loss: 0.5741 - val_accuracy: 0.8400 - 242ms/epoch - 5ms/step
Epoch 237/350
46/46 - 0s - loss: 0.5222 - accuracy: 0.8461 - val_loss: 0.4586 - val_accuracy: 0.8803 - 240ms/epoch - 5ms/step
Epoch 238/350
46/46 - 0s - loss: 0.5222 - accuracy: 0.8417 - val_loss: 0.5212 - val_accuracy: 0.8583 - 237ms/epoch - 5ms/step
Epoch 239/350
46/46 - 0s - loss: 0.5533 - accuracy: 0.8336 - val_loss: 0.4732 - val_accuracy: 0.8770 - 243ms/epoch - 5ms/step
Epoch 240/350
46/46 - 0s - loss: 0.5525 - accuracy: 0.8315 - val_loss: 0.4846 - val_accuracy: 0.8737 - 239ms/epoch - 5ms/step
Epoch 241/350
46/46 - 0s - loss: 0.5487 - accuracy: 0.8377 - val_loss: 0.4838 - val_accuracy: 0.8770 - 238ms/epoch - 5ms/step
Epoch 242/350
46/46 - 0s - loss: 0.5095 - accuracy: 0.8445 - val_loss: 0.4699 - val_accuracy: 0.8787 - 237ms/epoch - 5ms/step
Epoch 243/350
46/46 - 0s - loss: 0.5120 - accuracy: 0.8449 - val_loss: 0.4983 - val_accuracy: 0.8707 - 241ms/epoch - 5ms/step
Epoch 244/350
46/46 - 0s - loss: 0.5172 - accuracy: 0.8456 - val_loss: 0.4772 - val_accuracy: 0.8767 - 241ms/epoch - 5ms/step
Epoch 245/350
46/46 - 0s - loss: 0.4983 - accuracy: 0.8531 - val_loss: 0.4663 - val_accuracy: 0.8780 - 244ms/epoch - 5ms/step
Epoch 246/350
46/46 - 0s - loss: 0.5302 - accuracy: 0.8396 - val_loss: 0.4769 - val_accuracy: 0.8737 - 243ms/epoch - 5ms/step
Epoch 247/350
46/46 - 0s - loss: 0.5034 - accuracy: 0.8451 - val_loss: 0.4862 - val_accuracy: 0.8703 - 242ms/epoch - 5ms/step
Epoch 248/350
46/46 - 0s - loss: 0.5089 - accuracy: 0.8479 - val_loss: 0.4706 - val_accuracy: 0.8763 - 246ms/epoch - 5ms/step
Epoch 249/350
46/46 - 0s - loss: 0.4901 - accuracy: 0.8525 - val_loss: 0.4611 - val_accuracy: 0.8813 - 247ms/epoch - 5ms/step
Epoch 250/350
46/46 - 0s - loss: 0.5114 - accuracy: 0.8515 - val_loss: 0.4728 - val_accuracy: 0.8737 - 238ms/epoch - 5ms/step
Epoch 251/350
46/46 - 0s - loss: 0.4995 - accuracy: 0.8455 - val_loss: 0.4743 - val_accuracy: 0.8733 - 240ms/epoch - 5ms/step
Epoch 252/350
46/46 - 0s - loss: 0.5434 - accuracy: 0.8377 - val_loss: 0.4709 - val_accuracy: 0.8767 - 240ms/epoch - 5ms/step
Epoch 253/350
46/46 - 0s - loss: 0.4990 - accuracy: 0.8494 - val_loss: 0.4668 - val_accuracy: 0.8737 - 240ms/epoch - 5ms/step
Epoch 254/350
46/46 - 0s - loss: 0.5057 - accuracy: 0.8460 - val_loss: 0.4531 - val_accuracy: 0.8847 - 238ms/epoch - 5ms/step
Epoch 255/350
46/46 - 0s - loss: 0.5252 - accuracy: 0.8466 - val_loss: 0.4577 - val_accuracy: 0.8823 - 244ms/epoch - 5ms/step
Epoch 256/350
46/46 - 0s - loss: 0.5083 - accuracy: 0.8467 - val_loss: 0.4303 - val_accuracy: 0.8887 - 243ms/epoch - 5ms/step
Epoch 257/350
46/46 - 0s - loss: 0.4972 - accuracy: 0.8484 - val_loss: 0.4483 - val_accuracy: 0.8803 - 242ms/epoch - 5ms/step
Epoch 258/350
46/46 - 0s - loss: 0.5022 - accuracy: 0.8506 - val_loss: 0.4549 - val_accuracy: 0.8800 - 242ms/epoch - 5ms/step
Epoch 259/350
46/46 - 0s - loss: 0.4986 - accuracy: 0.8499 - val_loss: 0.4710 - val_accuracy: 0.8713 - 243ms/epoch - 5ms/step
Epoch 260/350
46/46 - 0s - loss: 0.5091 - accuracy: 0.8465 - val_loss: 0.4580 - val_accuracy: 0.8793 - 241ms/epoch - 5ms/step
Epoch 261/350
46/46 - 0s - loss: 0.5217 - accuracy: 0.8418 - val_loss: 0.4485 - val_accuracy: 0.8857 - 246ms/epoch - 5ms/step
Epoch 262/350
46/46 - 0s - loss: 0.4889 - accuracy: 0.8518 - val_loss: 0.4981 - val_accuracy: 0.8663 - 239ms/epoch - 5ms/step
Epoch 263/350
46/46 - 0s - loss: 0.5097 - accuracy: 0.8506 - val_loss: 0.4519 - val_accuracy: 0.8797 - 245ms/epoch - 5ms/step
Epoch 264/350
46/46 - 0s - loss: 0.4914 - accuracy: 0.8501 - val_loss: 0.4520 - val_accuracy: 0.8817 - 241ms/epoch - 5ms/step
Epoch 265/350
46/46 - 0s - loss: 0.4930 - accuracy: 0.8528 - val_loss: 0.4469 - val_accuracy: 0.8867 - 239ms/epoch - 5ms/step
Epoch 266/350
46/46 - 0s - loss: 0.4971 - accuracy: 0.8502 - val_loss: 0.4682 - val_accuracy: 0.8753 - 236ms/epoch - 5ms/step
Epoch 267/350
46/46 - 0s - loss: 0.4949 - accuracy: 0.8551 - val_loss: 0.4610 - val_accuracy: 0.8797 - 242ms/epoch - 5ms/step
Epoch 268/350
46/46 - 0s - loss: 0.4694 - accuracy: 0.8560 - val_loss: 0.4670 - val_accuracy: 0.8743 - 243ms/epoch - 5ms/step
Epoch 269/350
46/46 - 0s - loss: 0.5030 - accuracy: 0.8515 - val_loss: 0.4632 - val_accuracy: 0.8780 - 242ms/epoch - 5ms/step
Epoch 270/350
46/46 - 0s - loss: 0.4871 - accuracy: 0.8553 - val_loss: 0.4576 - val_accuracy: 0.8807 - 238ms/epoch - 5ms/step
Epoch 271/350
46/46 - 0s - loss: 0.4929 - accuracy: 0.8539 - val_loss: 0.4488 - val_accuracy: 0.8817 - 240ms/epoch - 5ms/step
Epoch 272/350
46/46 - 0s - loss: 0.4735 - accuracy: 0.8607 - val_loss: 0.4432 - val_accuracy: 0.8830 - 240ms/epoch - 5ms/step
Epoch 273/350
46/46 - 0s - loss: 0.4778 - accuracy: 0.8567 - val_loss: 0.5561 - val_accuracy: 0.8513 - 239ms/epoch - 5ms/step
Epoch 274/350
46/46 - 0s - loss: 0.5104 - accuracy: 0.8479 - val_loss: 0.4395 - val_accuracy: 0.8890 - 237ms/epoch - 5ms/step
Epoch 275/350
46/46 - 0s - loss: 0.4650 - accuracy: 0.8621 - val_loss: 0.4523 - val_accuracy: 0.8807 - 242ms/epoch - 5ms/step
Epoch 276/350
46/46 - 0s - loss: 0.5003 - accuracy: 0.8489 - val_loss: 0.5488 - val_accuracy: 0.8497 - 241ms/epoch - 5ms/step
Epoch 277/350
46/46 - 0s - loss: 0.5048 - accuracy: 0.8497 - val_loss: 0.4375 - val_accuracy: 0.8843 - 241ms/epoch - 5ms/step
Epoch 278/350
46/46 - 0s - loss: 0.4675 - accuracy: 0.8540 - val_loss: 0.4435 - val_accuracy: 0.8847 - 242ms/epoch - 5ms/step
Epoch 279/350
46/46 - 0s - loss: 0.4700 - accuracy: 0.8631 - val_loss: 0.4423 - val_accuracy: 0.8833 - 240ms/epoch - 5ms/step
Epoch 280/350
46/46 - 0s - loss: 0.4648 - accuracy: 0.8603 - val_loss: 0.4449 - val_accuracy: 0.8807 - 243ms/epoch - 5ms/step
Epoch 281/350
46/46 - 0s - loss: 0.4817 - accuracy: 0.8579 - val_loss: 0.4979 - val_accuracy: 0.8630 - 244ms/epoch - 5ms/step
Epoch 282/350
46/46 - 0s - loss: 0.4789 - accuracy: 0.8564 - val_loss: 0.4508 - val_accuracy: 0.8850 - 241ms/epoch - 5ms/step
Epoch 283/350
46/46 - 0s - loss: 0.4711 - accuracy: 0.8608 - val_loss: 0.4451 - val_accuracy: 0.8833 - 243ms/epoch - 5ms/step
Epoch 284/350
46/46 - 0s - loss: 0.4614 - accuracy: 0.8673 - val_loss: 0.4281 - val_accuracy: 0.8907 - 239ms/epoch - 5ms/step
Epoch 285/350
46/46 - 0s - loss: 0.4696 - accuracy: 0.8548 - val_loss: 0.4260 - val_accuracy: 0.8880 - 241ms/epoch - 5ms/step
Epoch 286/350
46/46 - 0s - loss: 0.4973 - accuracy: 0.8541 - val_loss: 0.4655 - val_accuracy: 0.8780 - 240ms/epoch - 5ms/step
Epoch 287/350
46/46 - 0s - loss: 0.4597 - accuracy: 0.8604 - val_loss: 0.4698 - val_accuracy: 0.8823 - 243ms/epoch - 5ms/step
Epoch 288/350
46/46 - 0s - loss: 0.4715 - accuracy: 0.8603 - val_loss: 0.4208 - val_accuracy: 0.8903 - 243ms/epoch - 5ms/step
Epoch 289/350
46/46 - 0s - loss: 0.4719 - accuracy: 0.8628 - val_loss: 0.4610 - val_accuracy: 0.8753 - 239ms/epoch - 5ms/step
Epoch 290/350
46/46 - 0s - loss: 0.4629 - accuracy: 0.8625 - val_loss: 0.4341 - val_accuracy: 0.8850 - 240ms/epoch - 5ms/step
Epoch 291/350
46/46 - 0s - loss: 0.4412 - accuracy: 0.8695 - val_loss: 0.4411 - val_accuracy: 0.8880 - 241ms/epoch - 5ms/step
Epoch 292/350
46/46 - 0s - loss: 0.4545 - accuracy: 0.8639 - val_loss: 0.4577 - val_accuracy: 0.8777 - 242ms/epoch - 5ms/step
Epoch 293/350
46/46 - 0s - loss: 0.4450 - accuracy: 0.8733 - val_loss: 0.4611 - val_accuracy: 0.8810 - 241ms/epoch - 5ms/step
Epoch 294/350
46/46 - 0s - loss: 0.4548 - accuracy: 0.8611 - val_loss: 0.4436 - val_accuracy: 0.8797 - 246ms/epoch - 5ms/step
Epoch 295/350
46/46 - 0s - loss: 0.4547 - accuracy: 0.8656 - val_loss: 0.4283 - val_accuracy: 0.8863 - 245ms/epoch - 5ms/step
Epoch 296/350
46/46 - 0s - loss: 0.4646 - accuracy: 0.8650 - val_loss: 0.4100 - val_accuracy: 0.8903 - 246ms/epoch - 5ms/step
Epoch 297/350
46/46 - 0s - loss: 0.4570 - accuracy: 0.8633 - val_loss: 0.4498 - val_accuracy: 0.8827 - 241ms/epoch - 5ms/step
Epoch 298/350
46/46 - 0s - loss: 0.4637 - accuracy: 0.8617 - val_loss: 0.4178 - val_accuracy: 0.8903 - 241ms/epoch - 5ms/step
Epoch 299/350
46/46 - 0s - loss: 0.4305 - accuracy: 0.8680 - val_loss: 0.4544 - val_accuracy: 0.8777 - 250ms/epoch - 5ms/step
Epoch 300/350
46/46 - 0s - loss: 0.4406 - accuracy: 0.8703 - val_loss: 0.4272 - val_accuracy: 0.8860 - 254ms/epoch - 6ms/step
Epoch 301/350
46/46 - 0s - loss: 0.4489 - accuracy: 0.8640 - val_loss: 0.4240 - val_accuracy: 0.8847 - 250ms/epoch - 5ms/step
Epoch 302/350
46/46 - 0s - loss: 0.4377 - accuracy: 0.8732 - val_loss: 0.4134 - val_accuracy: 0.8913 - 246ms/epoch - 5ms/step
Epoch 303/350
46/46 - 0s - loss: 0.4537 - accuracy: 0.8649 - val_loss: 0.4180 - val_accuracy: 0.8887 - 241ms/epoch - 5ms/step
Epoch 304/350
46/46 - 0s - loss: 0.4627 - accuracy: 0.8614 - val_loss: 0.4191 - val_accuracy: 0.8947 - 247ms/epoch - 5ms/step
Epoch 305/350
46/46 - 0s - loss: 0.4610 - accuracy: 0.8604 - val_loss: 0.4435 - val_accuracy: 0.8790 - 242ms/epoch - 5ms/step
Epoch 306/350
46/46 - 0s - loss: 0.4343 - accuracy: 0.8692 - val_loss: 0.4162 - val_accuracy: 0.8933 - 245ms/epoch - 5ms/step
Epoch 307/350
46/46 - 0s - loss: 0.4405 - accuracy: 0.8717 - val_loss: 0.4085 - val_accuracy: 0.8947 - 242ms/epoch - 5ms/step
Epoch 308/350
46/46 - 0s - loss: 0.4429 - accuracy: 0.8694 - val_loss: 0.4203 - val_accuracy: 0.8880 - 246ms/epoch - 5ms/step
Epoch 309/350
46/46 - 0s - loss: 0.4446 - accuracy: 0.8670 - val_loss: 0.4253 - val_accuracy: 0.8863 - 254ms/epoch - 6ms/step
Epoch 310/350
46/46 - 0s - loss: 0.4415 - accuracy: 0.8696 - val_loss: 0.4350 - val_accuracy: 0.8810 - 252ms/epoch - 5ms/step
Epoch 311/350
46/46 - 0s - loss: 0.4425 - accuracy: 0.8665 - val_loss: 0.4231 - val_accuracy: 0.8887 - 252ms/epoch - 5ms/step
Epoch 312/350
46/46 - 0s - loss: 0.4626 - accuracy: 0.8645 - val_loss: 0.4222 - val_accuracy: 0.8937 - 244ms/epoch - 5ms/step
Epoch 313/350
46/46 - 0s - loss: 0.4362 - accuracy: 0.8737 - val_loss: 0.4452 - val_accuracy: 0.8787 - 243ms/epoch - 5ms/step
Epoch 314/350
46/46 - 0s - loss: 0.4551 - accuracy: 0.8656 - val_loss: 0.4153 - val_accuracy: 0.8937 - 241ms/epoch - 5ms/step
Epoch 315/350
46/46 - 0s - loss: 0.4543 - accuracy: 0.8683 - val_loss: 0.4476 - val_accuracy: 0.8870 - 244ms/epoch - 5ms/step
Epoch 316/350
46/46 - 0s - loss: 0.4515 - accuracy: 0.8717 - val_loss: 0.4009 - val_accuracy: 0.8967 - 254ms/epoch - 6ms/step
Epoch 317/350
46/46 - 0s - loss: 0.4598 - accuracy: 0.8643 - val_loss: 0.4403 - val_accuracy: 0.8843 - 241ms/epoch - 5ms/step
Epoch 318/350
46/46 - 0s - loss: 0.4377 - accuracy: 0.8677 - val_loss: 0.4345 - val_accuracy: 0.8893 - 240ms/epoch - 5ms/step
Epoch 319/350
46/46 - 0s - loss: 0.4321 - accuracy: 0.8720 - val_loss: 0.4129 - val_accuracy: 0.8920 - 238ms/epoch - 5ms/step
Epoch 320/350
46/46 - 0s - loss: 0.4355 - accuracy: 0.8716 - val_loss: 0.4009 - val_accuracy: 0.8980 - 241ms/epoch - 5ms/step
Epoch 321/350
46/46 - 0s - loss: 0.4544 - accuracy: 0.8670 - val_loss: 0.4303 - val_accuracy: 0.8893 - 245ms/epoch - 5ms/step
Epoch 322/350
46/46 - 0s - loss: 0.4490 - accuracy: 0.8698 - val_loss: 0.4109 - val_accuracy: 0.8943 - 240ms/epoch - 5ms/step
Epoch 323/350
46/46 - 0s - loss: 0.4440 - accuracy: 0.8626 - val_loss: 0.5252 - val_accuracy: 0.8573 - 241ms/epoch - 5ms/step
Epoch 324/350
46/46 - 0s - loss: 0.4483 - accuracy: 0.8677 - val_loss: 0.4199 - val_accuracy: 0.8957 - 242ms/epoch - 5ms/step
Epoch 325/350
46/46 - 0s - loss: 0.4143 - accuracy: 0.8772 - val_loss: 0.4161 - val_accuracy: 0.8890 - 243ms/epoch - 5ms/step
Epoch 326/350
46/46 - 0s - loss: 0.4210 - accuracy: 0.8738 - val_loss: 0.4090 - val_accuracy: 0.8913 - 243ms/epoch - 5ms/step
Epoch 327/350
46/46 - 0s - loss: 0.4322 - accuracy: 0.8712 - val_loss: 0.4058 - val_accuracy: 0.8953 - 243ms/epoch - 5ms/step
Epoch 328/350
46/46 - 0s - loss: 0.4312 - accuracy: 0.8749 - val_loss: 0.4159 - val_accuracy: 0.8877 - 243ms/epoch - 5ms/step
Epoch 329/350
46/46 - 0s - loss: 0.4289 - accuracy: 0.8690 - val_loss: 0.4122 - val_accuracy: 0.8950 - 240ms/epoch - 5ms/step
Epoch 330/350
46/46 - 0s - loss: 0.4286 - accuracy: 0.8730 - val_loss: 0.3994 - val_accuracy: 0.8977 - 241ms/epoch - 5ms/step
Epoch 331/350
46/46 - 0s - loss: 0.4433 - accuracy: 0.8718 - val_loss: 0.4145 - val_accuracy: 0.8913 - 239ms/epoch - 5ms/step
Epoch 332/350
46/46 - 0s - loss: 0.4146 - accuracy: 0.8759 - val_loss: 0.4121 - val_accuracy: 0.8883 - 244ms/epoch - 5ms/step
Epoch 333/350
46/46 - 0s - loss: 0.4439 - accuracy: 0.8696 - val_loss: 0.4336 - val_accuracy: 0.8803 - 242ms/epoch - 5ms/step
Epoch 334/350
46/46 - 0s - loss: 0.4170 - accuracy: 0.8770 - val_loss: 0.4036 - val_accuracy: 0.8903 - 244ms/epoch - 5ms/step
Epoch 335/350
46/46 - 0s - loss: 0.4303 - accuracy: 0.8739 - val_loss: 0.4382 - val_accuracy: 0.8817 - 253ms/epoch - 5ms/step
Epoch 336/350
46/46 - 0s - loss: 0.4233 - accuracy: 0.8739 - val_loss: 0.3928 - val_accuracy: 0.8973 - 250ms/epoch - 5ms/step
Epoch 337/350
46/46 - 0s - loss: 0.4075 - accuracy: 0.8820 - val_loss: 0.3927 - val_accuracy: 0.8963 - 244ms/epoch - 5ms/step
Epoch 338/350
46/46 - 0s - loss: 0.4182 - accuracy: 0.8725 - val_loss: 0.4112 - val_accuracy: 0.8890 - 243ms/epoch - 5ms/step
Epoch 339/350
46/46 - 0s - loss: 0.4250 - accuracy: 0.8730 - val_loss: 0.4090 - val_accuracy: 0.8923 - 243ms/epoch - 5ms/step
Epoch 340/350
46/46 - 0s - loss: 0.4205 - accuracy: 0.8736 - val_loss: 0.4147 - val_accuracy: 0.8917 - 240ms/epoch - 5ms/step
Epoch 341/350
46/46 - 0s - loss: 0.4169 - accuracy: 0.8759 - val_loss: 0.4076 - val_accuracy: 0.8877 - 242ms/epoch - 5ms/step
Epoch 342/350
46/46 - 0s - loss: 0.4150 - accuracy: 0.8778 - val_loss: 0.3987 - val_accuracy: 0.8927 - 243ms/epoch - 5ms/step
Epoch 343/350
46/46 - 0s - loss: 0.4265 - accuracy: 0.8745 - val_loss: 0.4201 - val_accuracy: 0.8873 - 241ms/epoch - 5ms/step
Epoch 344/350
46/46 - 0s - loss: 0.4430 - accuracy: 0.8704 - val_loss: 0.4243 - val_accuracy: 0.8883 - 248ms/epoch - 5ms/step
Epoch 345/350
46/46 - 0s - loss: 0.4179 - accuracy: 0.8776 - val_loss: 0.4062 - val_accuracy: 0.8913 - 245ms/epoch - 5ms/step
Epoch 346/350
46/46 - 0s - loss: 0.4559 - accuracy: 0.8658 - val_loss: 0.4037 - val_accuracy: 0.8957 - 245ms/epoch - 5ms/step
Epoch 347/350
46/46 - 0s - loss: 0.4140 - accuracy: 0.8773 - val_loss: 0.4178 - val_accuracy: 0.8893 - 244ms/epoch - 5ms/step
Epoch 348/350
46/46 - 0s - loss: 0.4129 - accuracy: 0.8769 - val_loss: 0.4109 - val_accuracy: 0.8910 - 246ms/epoch - 5ms/step
Epoch 349/350
46/46 - 0s - loss: 0.4120 - accuracy: 0.8789 - val_loss: 0.4493 - val_accuracy: 0.8810 - 242ms/epoch - 5ms/step
Epoch 350/350
46/46 - 0s - loss: 0.4451 - accuracy: 0.8685 - val_loss: 0.4087 - val_accuracy: 0.8913 - 241ms/epoch - 5ms/step
94/94 - 1s - loss: 0.4087 - accuracy: 0.8913 - 694ms/epoch - 7ms/step
Baseline Error: 10.87%

model_31new.save('extraregularized31.h5') #Save model as results are promising

We can see that the model's performance now is excellent, we can see that near the end of the 300 epochs, there is still some sort of increase in accuracy, we can try to increase the epochs to 500 and see if we can get better results. However this is not a good idea as the runtime will increase by a lot for a small increase in accuracy, instead we can try to implement batch normalization for quicker convergence.

#Batch normalization

model_31new = models.Sequential([
    layers.Conv2D(32, (3, 3), activation='relu', input_shape=(31, 31, 1)),
    layers.MaxPooling2D((2, 2)),
    layers.Dropout(0.2),
    
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.MaxPooling2D((2, 2)),
    layers.Dropout(0.2),
    
    
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.MaxPooling2D((2, 2)),
    layers.Dropout(0.2),
    
    layers.Flatten(),
    
    layers.Dense(64, activation='relu', kernel_regularizer=regularizers.l2(0.001)),
    layers.Dropout(0.5),
    
    layers.BatchNormalization(), # add batch norm, i have tried other combinations, this seems to work the best with the least jumps in the data
    layers.Dense(15, activation='softmax')
])



model_31new.compile(optimizer='adam',
                loss= 'categorical_crossentropy',
                metrics=['accuracy'])

history = model_31new.fit(X_train31, train_labels, validation_data=(X_val31, validation_labels),
                    epochs=350, batch_size=200, verbose=2, class_weight = class_weight) #adjust epochs to 300

scores = model_31new.evaluate(X_val31, validation_labels, verbose=2)
print("Baseline Error: %.2f%%" % (100-scores[1]*100))
plot_history(history)
df31.loc[len(df31)] = ['NormalizedModel', history.history['accuracy'][-1], history.history['val_accuracy'][-1], history.history['loss'][-1], history.history['val_loss'][-1], history]

Epoch 1/350

2023-11-26 13:24:03.470223: E tensorflow/core/grappler/optimizers/meta_optimizer.cc:954] layout failed: INVALID_ARGUMENT: Size of values 0 does not match size of permutation 4 @ fanin shape insequential_11/dropout_7/dropout/SelectV2-2-TransposeNHWCToNCHW-LayoutOptimizer

46/46 - 3s - loss: 3.6186 - accuracy: 0.0757 - val_loss: 2.8064 - val_accuracy: 0.0667 - 3s/epoch - 61ms/step
Epoch 2/350
46/46 - 0s - loss: 3.2225 - accuracy: 0.1310 - val_loss: 2.8256 - val_accuracy: 0.1113 - 266ms/epoch - 6ms/step
Epoch 3/350
46/46 - 0s - loss: 2.9812 - accuracy: 0.1683 - val_loss: 2.8506 - val_accuracy: 0.1163 - 266ms/epoch - 6ms/step
Epoch 4/350
46/46 - 0s - loss: 2.8133 - accuracy: 0.1978 - val_loss: 2.8127 - val_accuracy: 0.1187 - 260ms/epoch - 6ms/step
Epoch 5/350
46/46 - 0s - loss: 2.6429 - accuracy: 0.2379 - val_loss: 2.7536 - val_accuracy: 0.1227 - 269ms/epoch - 6ms/step
Epoch 6/350
46/46 - 0s - loss: 2.4832 - accuracy: 0.2646 - val_loss: 2.6598 - val_accuracy: 0.1447 - 260ms/epoch - 6ms/step
Epoch 7/350
46/46 - 0s - loss: 2.3464 - accuracy: 0.2913 - val_loss: 2.5492 - val_accuracy: 0.1760 - 264ms/epoch - 6ms/step
Epoch 8/350
46/46 - 0s - loss: 2.2419 - accuracy: 0.3119 - val_loss: 2.4387 - val_accuracy: 0.2067 - 264ms/epoch - 6ms/step
Epoch 9/350
46/46 - 0s - loss: 2.1416 - accuracy: 0.3440 - val_loss: 2.2637 - val_accuracy: 0.2673 - 262ms/epoch - 6ms/step
Epoch 10/350
46/46 - 0s - loss: 2.0251 - accuracy: 0.3756 - val_loss: 2.0980 - val_accuracy: 0.3470 - 260ms/epoch - 6ms/step
Epoch 11/350
46/46 - 0s - loss: 1.9532 - accuracy: 0.3936 - val_loss: 2.0366 - val_accuracy: 0.3110 - 268ms/epoch - 6ms/step
Epoch 12/350
46/46 - 0s - loss: 1.8691 - accuracy: 0.4140 - val_loss: 1.8701 - val_accuracy: 0.4017 - 260ms/epoch - 6ms/step
Epoch 13/350
46/46 - 0s - loss: 1.8253 - accuracy: 0.4258 - val_loss: 1.8703 - val_accuracy: 0.3993 - 262ms/epoch - 6ms/step
Epoch 14/350
46/46 - 0s - loss: 1.7511 - accuracy: 0.4478 - val_loss: 1.6397 - val_accuracy: 0.4817 - 263ms/epoch - 6ms/step
Epoch 15/350
46/46 - 0s - loss: 1.6864 - accuracy: 0.4710 - val_loss: 1.6621 - val_accuracy: 0.4540 - 264ms/epoch - 6ms/step
Epoch 16/350
46/46 - 0s - loss: 1.6433 - accuracy: 0.4891 - val_loss: 1.5623 - val_accuracy: 0.5087 - 259ms/epoch - 6ms/step
Epoch 17/350
46/46 - 0s - loss: 1.5900 - accuracy: 0.5025 - val_loss: 1.4489 - val_accuracy: 0.5370 - 273ms/epoch - 6ms/step
Epoch 18/350
46/46 - 0s - loss: 1.5238 - accuracy: 0.5173 - val_loss: 1.3929 - val_accuracy: 0.5790 - 263ms/epoch - 6ms/step
Epoch 19/350
46/46 - 0s - loss: 1.4626 - accuracy: 0.5393 - val_loss: 1.2952 - val_accuracy: 0.5943 - 266ms/epoch - 6ms/step
Epoch 20/350
46/46 - 0s - loss: 1.4558 - accuracy: 0.5455 - val_loss: 1.2923 - val_accuracy: 0.6087 - 262ms/epoch - 6ms/step
Epoch 21/350
46/46 - 0s - loss: 1.3842 - accuracy: 0.5707 - val_loss: 1.2745 - val_accuracy: 0.6140 - 265ms/epoch - 6ms/step
Epoch 22/350
46/46 - 0s - loss: 1.3494 - accuracy: 0.5767 - val_loss: 1.2450 - val_accuracy: 0.6180 - 264ms/epoch - 6ms/step
Epoch 23/350
46/46 - 0s - loss: 1.2993 - accuracy: 0.5964 - val_loss: 1.1441 - val_accuracy: 0.6650 - 263ms/epoch - 6ms/step
Epoch 24/350
46/46 - 0s - loss: 1.2727 - accuracy: 0.6021 - val_loss: 1.1256 - val_accuracy: 0.6587 - 260ms/epoch - 6ms/step
Epoch 25/350
46/46 - 0s - loss: 1.2759 - accuracy: 0.5997 - val_loss: 1.2297 - val_accuracy: 0.6280 - 263ms/epoch - 6ms/step
Epoch 26/350
46/46 - 0s - loss: 1.2396 - accuracy: 0.6046 - val_loss: 1.0747 - val_accuracy: 0.6820 - 264ms/epoch - 6ms/step
Epoch 27/350
46/46 - 0s - loss: 1.1887 - accuracy: 0.6261 - val_loss: 1.0162 - val_accuracy: 0.6947 - 264ms/epoch - 6ms/step
Epoch 28/350
46/46 - 0s - loss: 1.1250 - accuracy: 0.6478 - val_loss: 0.9435 - val_accuracy: 0.7103 - 266ms/epoch - 6ms/step
Epoch 29/350
46/46 - 0s - loss: 1.1121 - accuracy: 0.6561 - val_loss: 0.9719 - val_accuracy: 0.7173 - 263ms/epoch - 6ms/step
Epoch 30/350
46/46 - 0s - loss: 1.0726 - accuracy: 0.6654 - val_loss: 0.9135 - val_accuracy: 0.7240 - 260ms/epoch - 6ms/step
Epoch 31/350
46/46 - 0s - loss: 1.0589 - accuracy: 0.6711 - val_loss: 0.9186 - val_accuracy: 0.7303 - 263ms/epoch - 6ms/step
Epoch 32/350
46/46 - 0s - loss: 1.0279 - accuracy: 0.6818 - val_loss: 0.8477 - val_accuracy: 0.7500 - 264ms/epoch - 6ms/step
Epoch 33/350
46/46 - 0s - loss: 1.0117 - accuracy: 0.6827 - val_loss: 0.9076 - val_accuracy: 0.7300 - 261ms/epoch - 6ms/step
Epoch 34/350
46/46 - 0s - loss: 1.0118 - accuracy: 0.6848 - val_loss: 0.7808 - val_accuracy: 0.7737 - 262ms/epoch - 6ms/step
Epoch 35/350
46/46 - 0s - loss: 0.9737 - accuracy: 0.7071 - val_loss: 0.8271 - val_accuracy: 0.7500 - 264ms/epoch - 6ms/step
Epoch 36/350
46/46 - 0s - loss: 0.9494 - accuracy: 0.7069 - val_loss: 0.7619 - val_accuracy: 0.7727 - 267ms/epoch - 6ms/step
Epoch 37/350
46/46 - 0s - loss: 0.9236 - accuracy: 0.7150 - val_loss: 0.8161 - val_accuracy: 0.7577 - 269ms/epoch - 6ms/step
Epoch 38/350
46/46 - 0s - loss: 0.9122 - accuracy: 0.7242 - val_loss: 0.7770 - val_accuracy: 0.7797 - 273ms/epoch - 6ms/step
Epoch 39/350
46/46 - 0s - loss: 0.9294 - accuracy: 0.7178 - val_loss: 0.7616 - val_accuracy: 0.7847 - 281ms/epoch - 6ms/step
Epoch 40/350
46/46 - 0s - loss: 0.8997 - accuracy: 0.7276 - val_loss: 0.8171 - val_accuracy: 0.7657 - 277ms/epoch - 6ms/step
Epoch 41/350
46/46 - 0s - loss: 0.8729 - accuracy: 0.7332 - val_loss: 0.6592 - val_accuracy: 0.8113 - 278ms/epoch - 6ms/step
Epoch 42/350
46/46 - 0s - loss: 0.8399 - accuracy: 0.7421 - val_loss: 0.6826 - val_accuracy: 0.8017 - 277ms/epoch - 6ms/step
Epoch 43/350
46/46 - 0s - loss: 0.8580 - accuracy: 0.7385 - val_loss: 0.7069 - val_accuracy: 0.7930 - 285ms/epoch - 6ms/step
Epoch 44/350
46/46 - 0s - loss: 0.8080 - accuracy: 0.7512 - val_loss: 0.7151 - val_accuracy: 0.7873 - 261ms/epoch - 6ms/step
Epoch 45/350
46/46 - 0s - loss: 0.8567 - accuracy: 0.7391 - val_loss: 0.7311 - val_accuracy: 0.7870 - 266ms/epoch - 6ms/step
Epoch 46/350
46/46 - 0s - loss: 0.7977 - accuracy: 0.7542 - val_loss: 0.6711 - val_accuracy: 0.8053 - 279ms/epoch - 6ms/step
Epoch 47/350
46/46 - 0s - loss: 0.7913 - accuracy: 0.7562 - val_loss: 0.6190 - val_accuracy: 0.8223 - 277ms/epoch - 6ms/step
Epoch 48/350
46/46 - 0s - loss: 0.7630 - accuracy: 0.7696 - val_loss: 0.6648 - val_accuracy: 0.8173 - 278ms/epoch - 6ms/step
Epoch 49/350
46/46 - 0s - loss: 0.7751 - accuracy: 0.7702 - val_loss: 0.6361 - val_accuracy: 0.8190 - 272ms/epoch - 6ms/step
Epoch 50/350
46/46 - 0s - loss: 0.7875 - accuracy: 0.7582 - val_loss: 0.6151 - val_accuracy: 0.8290 - 287ms/epoch - 6ms/step
Epoch 51/350
46/46 - 0s - loss: 0.7752 - accuracy: 0.7677 - val_loss: 0.6471 - val_accuracy: 0.8160 - 272ms/epoch - 6ms/step
Epoch 52/350
46/46 - 0s - loss: 0.7519 - accuracy: 0.7710 - val_loss: 0.6156 - val_accuracy: 0.8233 - 285ms/epoch - 6ms/step
Epoch 53/350
46/46 - 0s - loss: 0.7308 - accuracy: 0.7797 - val_loss: 0.6384 - val_accuracy: 0.8160 - 278ms/epoch - 6ms/step
Epoch 54/350
46/46 - 0s - loss: 0.6939 - accuracy: 0.7861 - val_loss: 0.5708 - val_accuracy: 0.8443 - 282ms/epoch - 6ms/step
Epoch 55/350
46/46 - 0s - loss: 0.7007 - accuracy: 0.7905 - val_loss: 0.5640 - val_accuracy: 0.8483 - 273ms/epoch - 6ms/step
Epoch 56/350
46/46 - 0s - loss: 0.6928 - accuracy: 0.7902 - val_loss: 0.5875 - val_accuracy: 0.8307 - 263ms/epoch - 6ms/step
Epoch 57/350
46/46 - 0s - loss: 0.6450 - accuracy: 0.8065 - val_loss: 0.5644 - val_accuracy: 0.8370 - 266ms/epoch - 6ms/step
Epoch 58/350
46/46 - 0s - loss: 0.6533 - accuracy: 0.8013 - val_loss: 0.5848 - val_accuracy: 0.8463 - 278ms/epoch - 6ms/step
Epoch 59/350
46/46 - 0s - loss: 0.6985 - accuracy: 0.7933 - val_loss: 0.5802 - val_accuracy: 0.8377 - 272ms/epoch - 6ms/step
Epoch 60/350
46/46 - 0s - loss: 0.6507 - accuracy: 0.8035 - val_loss: 0.5557 - val_accuracy: 0.8433 - 266ms/epoch - 6ms/step
Epoch 61/350
46/46 - 0s - loss: 0.6414 - accuracy: 0.8028 - val_loss: 0.5570 - val_accuracy: 0.8440 - 263ms/epoch - 6ms/step
Epoch 62/350
46/46 - 0s - loss: 0.6392 - accuracy: 0.8037 - val_loss: 0.5160 - val_accuracy: 0.8573 - 265ms/epoch - 6ms/step
Epoch 63/350
46/46 - 0s - loss: 0.6297 - accuracy: 0.8090 - val_loss: 0.5461 - val_accuracy: 0.8520 - 267ms/epoch - 6ms/step
Epoch 64/350
46/46 - 0s - loss: 0.6606 - accuracy: 0.8047 - val_loss: 0.5390 - val_accuracy: 0.8497 - 260ms/epoch - 6ms/step
Epoch 65/350
46/46 - 0s - loss: 0.6932 - accuracy: 0.7977 - val_loss: 0.5616 - val_accuracy: 0.8373 - 264ms/epoch - 6ms/step
Epoch 66/350
46/46 - 0s - loss: 0.6110 - accuracy: 0.8150 - val_loss: 0.5209 - val_accuracy: 0.8623 - 262ms/epoch - 6ms/step
Epoch 67/350
46/46 - 0s - loss: 0.6179 - accuracy: 0.8163 - val_loss: 0.5357 - val_accuracy: 0.8553 - 255ms/epoch - 6ms/step
Epoch 68/350
46/46 - 0s - loss: 0.6124 - accuracy: 0.8190 - val_loss: 0.5707 - val_accuracy: 0.8390 - 262ms/epoch - 6ms/step
Epoch 69/350
46/46 - 0s - loss: 0.6006 - accuracy: 0.8211 - val_loss: 0.5219 - val_accuracy: 0.8590 - 264ms/epoch - 6ms/step
Epoch 70/350
46/46 - 0s - loss: 0.6008 - accuracy: 0.8175 - val_loss: 0.5472 - val_accuracy: 0.8447 - 276ms/epoch - 6ms/step
Epoch 71/350
46/46 - 0s - loss: 0.6240 - accuracy: 0.8127 - val_loss: 0.4887 - val_accuracy: 0.8617 - 273ms/epoch - 6ms/step
Epoch 72/350
46/46 - 0s - loss: 0.5801 - accuracy: 0.8304 - val_loss: 0.4981 - val_accuracy: 0.8600 - 272ms/epoch - 6ms/step
Epoch 73/350
46/46 - 0s - loss: 0.5820 - accuracy: 0.8271 - val_loss: 0.5487 - val_accuracy: 0.8477 - 271ms/epoch - 6ms/step
Epoch 74/350
46/46 - 0s - loss: 0.5766 - accuracy: 0.8269 - val_loss: 0.5035 - val_accuracy: 0.8677 - 263ms/epoch - 6ms/step
Epoch 75/350
46/46 - 0s - loss: 0.5678 - accuracy: 0.8356 - val_loss: 0.4847 - val_accuracy: 0.8633 - 264ms/epoch - 6ms/step
Epoch 76/350
46/46 - 0s - loss: 0.5764 - accuracy: 0.8267 - val_loss: 0.4969 - val_accuracy: 0.8613 - 261ms/epoch - 6ms/step
Epoch 77/350
46/46 - 0s - loss: 0.5761 - accuracy: 0.8275 - val_loss: 0.5135 - val_accuracy: 0.8563 - 263ms/epoch - 6ms/step
Epoch 78/350
46/46 - 0s - loss: 0.5680 - accuracy: 0.8309 - val_loss: 0.5291 - val_accuracy: 0.8547 - 259ms/epoch - 6ms/step
Epoch 79/350
46/46 - 0s - loss: 0.6102 - accuracy: 0.8187 - val_loss: 0.5424 - val_accuracy: 0.8457 - 258ms/epoch - 6ms/step
Epoch 80/350
46/46 - 0s - loss: 0.5956 - accuracy: 0.8263 - val_loss: 0.5895 - val_accuracy: 0.8447 - 261ms/epoch - 6ms/step
Epoch 81/350
46/46 - 0s - loss: 0.5837 - accuracy: 0.8233 - val_loss: 0.4988 - val_accuracy: 0.8587 - 262ms/epoch - 6ms/step
Epoch 82/350
46/46 - 0s - loss: 0.5211 - accuracy: 0.8401 - val_loss: 0.4909 - val_accuracy: 0.8647 - 263ms/epoch - 6ms/step
Epoch 83/350
46/46 - 0s - loss: 0.5529 - accuracy: 0.8350 - val_loss: 0.4636 - val_accuracy: 0.8773 - 276ms/epoch - 6ms/step
Epoch 84/350
46/46 - 0s - loss: 0.5448 - accuracy: 0.8356 - val_loss: 0.4849 - val_accuracy: 0.8630 - 285ms/epoch - 6ms/step
Epoch 85/350
46/46 - 0s - loss: 0.5352 - accuracy: 0.8424 - val_loss: 0.4724 - val_accuracy: 0.8693 - 265ms/epoch - 6ms/step
Epoch 86/350
46/46 - 0s - loss: 0.5768 - accuracy: 0.8263 - val_loss: 0.4696 - val_accuracy: 0.8657 - 259ms/epoch - 6ms/step
Epoch 87/350
46/46 - 0s - loss: 0.5493 - accuracy: 0.8405 - val_loss: 0.5156 - val_accuracy: 0.8677 - 262ms/epoch - 6ms/step
Epoch 88/350
46/46 - 0s - loss: 0.5374 - accuracy: 0.8405 - val_loss: 0.4564 - val_accuracy: 0.8753 - 259ms/epoch - 6ms/step
Epoch 89/350
46/46 - 0s - loss: 0.5292 - accuracy: 0.8420 - val_loss: 0.4756 - val_accuracy: 0.8690 - 263ms/epoch - 6ms/step
Epoch 90/350
46/46 - 0s - loss: 0.5109 - accuracy: 0.8479 - val_loss: 0.4803 - val_accuracy: 0.8663 - 262ms/epoch - 6ms/step
Epoch 91/350
46/46 - 0s - loss: 0.4832 - accuracy: 0.8549 - val_loss: 0.4472 - val_accuracy: 0.8770 - 264ms/epoch - 6ms/step
Epoch 92/350
46/46 - 0s - loss: 0.5517 - accuracy: 0.8346 - val_loss: 0.4672 - val_accuracy: 0.8733 - 260ms/epoch - 6ms/step
Epoch 93/350
46/46 - 0s - loss: 0.5339 - accuracy: 0.8383 - val_loss: 0.4782 - val_accuracy: 0.8707 - 260ms/epoch - 6ms/step
Epoch 94/350
46/46 - 0s - loss: 0.5006 - accuracy: 0.8557 - val_loss: 0.4433 - val_accuracy: 0.8763 - 260ms/epoch - 6ms/step
Epoch 95/350
46/46 - 0s - loss: 0.5155 - accuracy: 0.8487 - val_loss: 0.4633 - val_accuracy: 0.8743 - 284ms/epoch - 6ms/step
Epoch 96/350
46/46 - 0s - loss: 0.5190 - accuracy: 0.8475 - val_loss: 0.4290 - val_accuracy: 0.8823 - 258ms/epoch - 6ms/step
Epoch 97/350
46/46 - 0s - loss: 0.5114 - accuracy: 0.8487 - val_loss: 0.4497 - val_accuracy: 0.8727 - 259ms/epoch - 6ms/step
Epoch 98/350
46/46 - 0s - loss: 0.4828 - accuracy: 0.8543 - val_loss: 0.4127 - val_accuracy: 0.8863 - 261ms/epoch - 6ms/step
Epoch 99/350
46/46 - 0s - loss: 0.4861 - accuracy: 0.8535 - val_loss: 0.4453 - val_accuracy: 0.8800 - 266ms/epoch - 6ms/step
Epoch 100/350
46/46 - 0s - loss: 0.4744 - accuracy: 0.8543 - val_loss: 0.5469 - val_accuracy: 0.8510 - 270ms/epoch - 6ms/step
Epoch 101/350
46/46 - 0s - loss: 0.4869 - accuracy: 0.8554 - val_loss: 0.4297 - val_accuracy: 0.8847 - 258ms/epoch - 6ms/step
Epoch 102/350
46/46 - 0s - loss: 0.4697 - accuracy: 0.8598 - val_loss: 0.4597 - val_accuracy: 0.8810 - 267ms/epoch - 6ms/step
Epoch 103/350
46/46 - 0s - loss: 0.4977 - accuracy: 0.8533 - val_loss: 0.4290 - val_accuracy: 0.8850 - 264ms/epoch - 6ms/step
Epoch 104/350
46/46 - 0s - loss: 0.4758 - accuracy: 0.8551 - val_loss: 0.4216 - val_accuracy: 0.8890 - 263ms/epoch - 6ms/step
Epoch 105/350
46/46 - 0s - loss: 0.4579 - accuracy: 0.8645 - val_loss: 0.4149 - val_accuracy: 0.8870 - 264ms/epoch - 6ms/step
Epoch 106/350
46/46 - 0s - loss: 0.4617 - accuracy: 0.8628 - val_loss: 0.4330 - val_accuracy: 0.8837 - 261ms/epoch - 6ms/step
Epoch 107/350
46/46 - 0s - loss: 0.5064 - accuracy: 0.8514 - val_loss: 0.4291 - val_accuracy: 0.8870 - 262ms/epoch - 6ms/step
Epoch 108/350
46/46 - 0s - loss: 0.4745 - accuracy: 0.8578 - val_loss: 0.4891 - val_accuracy: 0.8703 - 255ms/epoch - 6ms/step
Epoch 109/350
46/46 - 0s - loss: 0.4779 - accuracy: 0.8648 - val_loss: 0.4175 - val_accuracy: 0.8890 - 260ms/epoch - 6ms/step
Epoch 110/350
46/46 - 0s - loss: 0.4796 - accuracy: 0.8623 - val_loss: 0.4704 - val_accuracy: 0.8693 - 275ms/epoch - 6ms/step
Epoch 111/350
46/46 - 0s - loss: 0.4575 - accuracy: 0.8604 - val_loss: 0.4218 - val_accuracy: 0.8867 - 263ms/epoch - 6ms/step
Epoch 112/350
46/46 - 0s - loss: 0.4762 - accuracy: 0.8591 - val_loss: 0.4252 - val_accuracy: 0.8847 - 258ms/epoch - 6ms/step
Epoch 113/350
46/46 - 0s - loss: 0.4554 - accuracy: 0.8655 - val_loss: 0.4239 - val_accuracy: 0.8840 - 259ms/epoch - 6ms/step
Epoch 114/350
46/46 - 0s - loss: 0.4659 - accuracy: 0.8611 - val_loss: 0.4570 - val_accuracy: 0.8803 - 260ms/epoch - 6ms/step
Epoch 115/350
46/46 - 0s - loss: 0.4510 - accuracy: 0.8664 - val_loss: 0.4409 - val_accuracy: 0.8847 - 257ms/epoch - 6ms/step
Epoch 116/350
46/46 - 0s - loss: 0.4619 - accuracy: 0.8614 - val_loss: 0.4310 - val_accuracy: 0.8843 - 257ms/epoch - 6ms/step
Epoch 117/350
46/46 - 0s - loss: 0.4713 - accuracy: 0.8646 - val_loss: 0.4075 - val_accuracy: 0.8920 - 263ms/epoch - 6ms/step
Epoch 118/350
46/46 - 0s - loss: 0.4495 - accuracy: 0.8642 - val_loss: 0.4761 - val_accuracy: 0.8720 - 255ms/epoch - 6ms/step
Epoch 119/350
46/46 - 0s - loss: 0.4626 - accuracy: 0.8651 - val_loss: 0.4053 - val_accuracy: 0.8877 - 262ms/epoch - 6ms/step
Epoch 120/350
46/46 - 0s - loss: 0.4464 - accuracy: 0.8708 - val_loss: 0.4233 - val_accuracy: 0.8883 - 257ms/epoch - 6ms/step
Epoch 121/350
46/46 - 0s - loss: 0.4337 - accuracy: 0.8720 - val_loss: 0.4256 - val_accuracy: 0.8827 - 260ms/epoch - 6ms/step
Epoch 122/350
46/46 - 0s - loss: 0.4488 - accuracy: 0.8711 - val_loss: 0.4240 - val_accuracy: 0.8860 - 263ms/epoch - 6ms/step
Epoch 123/350
46/46 - 0s - loss: 0.4668 - accuracy: 0.8634 - val_loss: 0.4189 - val_accuracy: 0.8873 - 258ms/epoch - 6ms/step
Epoch 124/350
46/46 - 0s - loss: 0.4609 - accuracy: 0.8654 - val_loss: 0.4133 - val_accuracy: 0.8927 - 254ms/epoch - 6ms/step
Epoch 125/350
46/46 - 0s - loss: 0.4385 - accuracy: 0.8704 - val_loss: 0.4730 - val_accuracy: 0.8733 - 261ms/epoch - 6ms/step
Epoch 126/350
46/46 - 0s - loss: 0.4435 - accuracy: 0.8720 - val_loss: 0.4021 - val_accuracy: 0.8900 - 265ms/epoch - 6ms/step
Epoch 127/350
46/46 - 0s - loss: 0.4545 - accuracy: 0.8661 - val_loss: 0.4131 - val_accuracy: 0.8897 - 260ms/epoch - 6ms/step
Epoch 128/350
46/46 - 0s - loss: 0.4372 - accuracy: 0.8730 - val_loss: 0.4424 - val_accuracy: 0.8850 - 261ms/epoch - 6ms/step
Epoch 129/350
46/46 - 0s - loss: 0.4578 - accuracy: 0.8667 - val_loss: 0.4032 - val_accuracy: 0.8933 - 269ms/epoch - 6ms/step
Epoch 130/350
46/46 - 0s - loss: 0.4086 - accuracy: 0.8798 - val_loss: 0.4092 - val_accuracy: 0.8903 - 259ms/epoch - 6ms/step
Epoch 131/350
46/46 - 0s - loss: 0.4418 - accuracy: 0.8703 - val_loss: 0.4652 - val_accuracy: 0.8790 - 257ms/epoch - 6ms/step
Epoch 132/350
46/46 - 0s - loss: 0.4344 - accuracy: 0.8728 - val_loss: 0.4780 - val_accuracy: 0.8737 - 262ms/epoch - 6ms/step
Epoch 133/350
46/46 - 0s - loss: 0.4089 - accuracy: 0.8808 - val_loss: 0.4567 - val_accuracy: 0.8767 - 261ms/epoch - 6ms/step
Epoch 134/350
46/46 - 0s - loss: 0.4437 - accuracy: 0.8708 - val_loss: 0.4129 - val_accuracy: 0.8893 - 257ms/epoch - 6ms/step
Epoch 135/350
46/46 - 0s - loss: 0.4421 - accuracy: 0.8726 - val_loss: 0.4518 - val_accuracy: 0.8847 - 257ms/epoch - 6ms/step
Epoch 136/350
46/46 - 0s - loss: 0.4155 - accuracy: 0.8799 - val_loss: 0.4168 - val_accuracy: 0.8897 - 262ms/epoch - 6ms/step
Epoch 137/350
46/46 - 0s - loss: 0.4062 - accuracy: 0.8810 - val_loss: 0.3879 - val_accuracy: 0.8967 - 260ms/epoch - 6ms/step
Epoch 138/350
46/46 - 0s - loss: 0.4131 - accuracy: 0.8795 - val_loss: 0.3896 - val_accuracy: 0.8997 - 261ms/epoch - 6ms/step
Epoch 139/350
46/46 - 0s - loss: 0.4178 - accuracy: 0.8802 - val_loss: 0.4348 - val_accuracy: 0.8790 - 257ms/epoch - 6ms/step
Epoch 140/350
46/46 - 0s - loss: 0.4340 - accuracy: 0.8780 - val_loss: 0.4241 - val_accuracy: 0.8903 - 260ms/epoch - 6ms/step
Epoch 141/350
46/46 - 0s - loss: 0.4030 - accuracy: 0.8831 - val_loss: 0.4326 - val_accuracy: 0.8873 - 259ms/epoch - 6ms/step
Epoch 142/350
46/46 - 0s - loss: 0.4193 - accuracy: 0.8756 - val_loss: 0.4401 - val_accuracy: 0.8817 - 256ms/epoch - 6ms/step
Epoch 143/350
46/46 - 0s - loss: 0.4104 - accuracy: 0.8767 - val_loss: 0.4116 - val_accuracy: 0.8947 - 257ms/epoch - 6ms/step
Epoch 144/350
46/46 - 0s - loss: 0.4114 - accuracy: 0.8808 - val_loss: 0.4063 - val_accuracy: 0.8890 - 257ms/epoch - 6ms/step
Epoch 145/350
46/46 - 0s - loss: 0.4200 - accuracy: 0.8751 - val_loss: 0.4228 - val_accuracy: 0.8960 - 257ms/epoch - 6ms/step
Epoch 146/350
46/46 - 0s - loss: 0.4193 - accuracy: 0.8764 - val_loss: 0.4261 - val_accuracy: 0.8900 - 258ms/epoch - 6ms/step
Epoch 147/350
46/46 - 0s - loss: 0.4083 - accuracy: 0.8785 - val_loss: 0.3998 - val_accuracy: 0.8960 - 263ms/epoch - 6ms/step
Epoch 148/350
46/46 - 0s - loss: 0.4102 - accuracy: 0.8794 - val_loss: 0.4381 - val_accuracy: 0.8810 - 260ms/epoch - 6ms/step
Epoch 149/350
46/46 - 0s - loss: 0.4371 - accuracy: 0.8769 - val_loss: 0.4038 - val_accuracy: 0.8947 - 277ms/epoch - 6ms/step
Epoch 150/350
46/46 - 0s - loss: 0.4210 - accuracy: 0.8743 - val_loss: 0.4274 - val_accuracy: 0.8833 - 261ms/epoch - 6ms/step
Epoch 151/350
46/46 - 0s - loss: 0.4152 - accuracy: 0.8792 - val_loss: 0.4303 - val_accuracy: 0.8847 - 269ms/epoch - 6ms/step
Epoch 152/350
46/46 - 0s - loss: 0.4750 - accuracy: 0.8621 - val_loss: 0.4072 - val_accuracy: 0.8903 - 263ms/epoch - 6ms/step
Epoch 153/350
46/46 - 0s - loss: 0.4174 - accuracy: 0.8772 - val_loss: 0.4040 - val_accuracy: 0.8947 - 263ms/epoch - 6ms/step
Epoch 154/350
46/46 - 0s - loss: 0.4090 - accuracy: 0.8798 - val_loss: 0.4877 - val_accuracy: 0.8703 - 260ms/epoch - 6ms/step
Epoch 155/350
46/46 - 0s - loss: 0.4592 - accuracy: 0.8649 - val_loss: 0.4289 - val_accuracy: 0.8813 - 266ms/epoch - 6ms/step
Epoch 156/350
46/46 - 0s - loss: 0.4070 - accuracy: 0.8825 - val_loss: 0.4091 - val_accuracy: 0.8923 - 261ms/epoch - 6ms/step
Epoch 157/350
46/46 - 0s - loss: 0.4106 - accuracy: 0.8776 - val_loss: 0.4105 - val_accuracy: 0.8923 - 270ms/epoch - 6ms/step
Epoch 158/350
46/46 - 0s - loss: 0.3932 - accuracy: 0.8871 - val_loss: 0.3968 - val_accuracy: 0.8937 - 263ms/epoch - 6ms/step
Epoch 159/350
46/46 - 0s - loss: 0.3945 - accuracy: 0.8846 - val_loss: 0.3977 - val_accuracy: 0.8943 - 269ms/epoch - 6ms/step
Epoch 160/350
46/46 - 0s - loss: 0.3881 - accuracy: 0.8893 - val_loss: 0.4059 - val_accuracy: 0.8867 - 262ms/epoch - 6ms/step
Epoch 161/350
46/46 - 0s - loss: 0.3781 - accuracy: 0.8846 - val_loss: 0.3824 - val_accuracy: 0.9060 - 266ms/epoch - 6ms/step
Epoch 162/350
46/46 - 0s - loss: 0.3959 - accuracy: 0.8859 - val_loss: 0.4257 - val_accuracy: 0.8903 - 259ms/epoch - 6ms/step
Epoch 163/350
46/46 - 0s - loss: 0.3914 - accuracy: 0.8888 - val_loss: 0.4240 - val_accuracy: 0.8853 - 261ms/epoch - 6ms/step
Epoch 164/350
46/46 - 0s - loss: 0.3868 - accuracy: 0.8905 - val_loss: 0.3882 - val_accuracy: 0.8983 - 264ms/epoch - 6ms/step
Epoch 165/350
46/46 - 0s - loss: 0.3931 - accuracy: 0.8865 - val_loss: 0.5230 - val_accuracy: 0.8643 - 267ms/epoch - 6ms/step
Epoch 166/350
46/46 - 0s - loss: 0.4029 - accuracy: 0.8808 - val_loss: 0.3885 - val_accuracy: 0.8960 - 270ms/epoch - 6ms/step
Epoch 167/350
46/46 - 0s - loss: 0.3794 - accuracy: 0.8897 - val_loss: 0.3853 - val_accuracy: 0.9027 - 267ms/epoch - 6ms/step
Epoch 168/350
46/46 - 0s - loss: 0.3729 - accuracy: 0.8919 - val_loss: 0.3798 - val_accuracy: 0.9077 - 272ms/epoch - 6ms/step
Epoch 169/350
46/46 - 0s - loss: 0.3757 - accuracy: 0.8924 - val_loss: 0.4181 - val_accuracy: 0.8920 - 271ms/epoch - 6ms/step
Epoch 170/350
46/46 - 0s - loss: 0.3822 - accuracy: 0.8866 - val_loss: 0.4021 - val_accuracy: 0.8960 - 271ms/epoch - 6ms/step
Epoch 171/350
46/46 - 0s - loss: 0.3655 - accuracy: 0.8947 - val_loss: 0.3848 - val_accuracy: 0.8950 - 274ms/epoch - 6ms/step
Epoch 172/350
46/46 - 0s - loss: 0.3708 - accuracy: 0.8875 - val_loss: 0.4120 - val_accuracy: 0.8980 - 279ms/epoch - 6ms/step
Epoch 173/350
46/46 - 0s - loss: 0.3780 - accuracy: 0.8909 - val_loss: 0.4001 - val_accuracy: 0.9010 - 275ms/epoch - 6ms/step
Epoch 174/350
46/46 - 0s - loss: 0.3683 - accuracy: 0.8931 - val_loss: 0.4558 - val_accuracy: 0.8863 - 268ms/epoch - 6ms/step
Epoch 175/350
46/46 - 0s - loss: 0.4167 - accuracy: 0.8790 - val_loss: 0.3846 - val_accuracy: 0.9007 - 267ms/epoch - 6ms/step
Epoch 176/350
46/46 - 0s - loss: 0.3788 - accuracy: 0.8897 - val_loss: 0.4310 - val_accuracy: 0.8877 - 258ms/epoch - 6ms/step
Epoch 177/350
46/46 - 0s - loss: 0.3849 - accuracy: 0.8865 - val_loss: 0.3781 - val_accuracy: 0.9000 - 269ms/epoch - 6ms/step
Epoch 178/350
46/46 - 0s - loss: 0.4045 - accuracy: 0.8874 - val_loss: 0.4145 - val_accuracy: 0.8930 - 262ms/epoch - 6ms/step
Epoch 179/350
46/46 - 0s - loss: 0.3617 - accuracy: 0.8938 - val_loss: 0.3910 - val_accuracy: 0.9027 - 261ms/epoch - 6ms/step
Epoch 180/350
46/46 - 0s - loss: 0.4019 - accuracy: 0.8810 - val_loss: 0.3767 - val_accuracy: 0.9013 - 259ms/epoch - 6ms/step
Epoch 181/350
46/46 - 0s - loss: 0.3762 - accuracy: 0.8892 - val_loss: 0.4118 - val_accuracy: 0.8883 - 263ms/epoch - 6ms/step
Epoch 182/350
46/46 - 0s - loss: 0.3694 - accuracy: 0.8931 - val_loss: 0.3936 - val_accuracy: 0.8940 - 275ms/epoch - 6ms/step
Epoch 183/350
46/46 - 0s - loss: 0.3755 - accuracy: 0.8874 - val_loss: 0.4147 - val_accuracy: 0.8900 - 261ms/epoch - 6ms/step
Epoch 184/350
46/46 - 0s - loss: 0.3855 - accuracy: 0.8918 - val_loss: 0.3992 - val_accuracy: 0.8940 - 261ms/epoch - 6ms/step
Epoch 185/350
46/46 - 0s - loss: 0.3671 - accuracy: 0.8968 - val_loss: 0.3854 - val_accuracy: 0.8983 - 263ms/epoch - 6ms/step
Epoch 186/350
46/46 - 0s - loss: 0.3668 - accuracy: 0.8952 - val_loss: 0.3855 - val_accuracy: 0.9003 - 260ms/epoch - 6ms/step
Epoch 187/350
46/46 - 0s - loss: 0.3730 - accuracy: 0.8959 - val_loss: 0.3700 - val_accuracy: 0.8987 - 259ms/epoch - 6ms/step
Epoch 188/350
46/46 - 0s - loss: 0.3608 - accuracy: 0.8989 - val_loss: 0.3953 - val_accuracy: 0.8953 - 261ms/epoch - 6ms/step
Epoch 189/350
46/46 - 0s - loss: 0.3608 - accuracy: 0.8950 - val_loss: 0.3754 - val_accuracy: 0.8997 - 265ms/epoch - 6ms/step
Epoch 190/350
46/46 - 0s - loss: 0.3780 - accuracy: 0.8957 - val_loss: 0.3857 - val_accuracy: 0.8943 - 262ms/epoch - 6ms/step
Epoch 191/350
46/46 - 0s - loss: 0.3689 - accuracy: 0.8954 - val_loss: 0.4086 - val_accuracy: 0.8923 - 256ms/epoch - 6ms/step
Epoch 192/350
46/46 - 0s - loss: 0.3898 - accuracy: 0.8870 - val_loss: 0.4272 - val_accuracy: 0.8917 - 263ms/epoch - 6ms/step
Epoch 193/350
46/46 - 0s - loss: 0.3800 - accuracy: 0.8897 - val_loss: 0.3914 - val_accuracy: 0.8937 - 263ms/epoch - 6ms/step
Epoch 194/350
46/46 - 0s - loss: 0.3879 - accuracy: 0.8889 - val_loss: 0.3835 - val_accuracy: 0.8993 - 262ms/epoch - 6ms/step
Epoch 195/350
46/46 - 0s - loss: 0.3633 - accuracy: 0.8961 - val_loss: 0.3742 - val_accuracy: 0.9030 - 261ms/epoch - 6ms/step
Epoch 196/350
46/46 - 0s - loss: 0.3710 - accuracy: 0.8955 - val_loss: 0.3768 - val_accuracy: 0.9020 - 263ms/epoch - 6ms/step
Epoch 197/350
46/46 - 0s - loss: 0.3688 - accuracy: 0.8910 - val_loss: 0.3625 - val_accuracy: 0.9103 - 260ms/epoch - 6ms/step
Epoch 198/350
46/46 - 0s - loss: 0.3612 - accuracy: 0.8960 - val_loss: 0.4077 - val_accuracy: 0.8930 - 263ms/epoch - 6ms/step
Epoch 199/350
46/46 - 0s - loss: 0.3702 - accuracy: 0.8949 - val_loss: 0.4507 - val_accuracy: 0.8807 - 273ms/epoch - 6ms/step
Epoch 200/350
46/46 - 0s - loss: 0.3700 - accuracy: 0.8924 - val_loss: 0.3987 - val_accuracy: 0.8943 - 265ms/epoch - 6ms/step
Epoch 201/350
46/46 - 0s - loss: 0.3265 - accuracy: 0.9053 - val_loss: 0.3953 - val_accuracy: 0.8987 - 258ms/epoch - 6ms/step
Epoch 202/350
46/46 - 0s - loss: 0.3349 - accuracy: 0.9039 - val_loss: 0.3983 - val_accuracy: 0.8963 - 278ms/epoch - 6ms/step
Epoch 203/350
46/46 - 0s - loss: 0.3364 - accuracy: 0.9035 - val_loss: 0.3795 - val_accuracy: 0.9060 - 288ms/epoch - 6ms/step
Epoch 204/350
46/46 - 0s - loss: 0.3433 - accuracy: 0.8985 - val_loss: 0.4432 - val_accuracy: 0.8850 - 279ms/epoch - 6ms/step
Epoch 205/350
46/46 - 0s - loss: 0.3537 - accuracy: 0.8954 - val_loss: 0.3842 - val_accuracy: 0.9037 - 267ms/epoch - 6ms/step
Epoch 206/350
46/46 - 0s - loss: 0.3829 - accuracy: 0.8920 - val_loss: 0.4297 - val_accuracy: 0.8910 - 256ms/epoch - 6ms/step
Epoch 207/350
46/46 - 0s - loss: 0.3404 - accuracy: 0.9027 - val_loss: 0.3962 - val_accuracy: 0.8913 - 262ms/epoch - 6ms/step
Epoch 208/350
46/46 - 0s - loss: 0.3508 - accuracy: 0.8978 - val_loss: 0.4120 - val_accuracy: 0.8873 - 260ms/epoch - 6ms/step
Epoch 209/350
46/46 - 0s - loss: 0.3404 - accuracy: 0.9002 - val_loss: 0.3919 - val_accuracy: 0.9033 - 262ms/epoch - 6ms/step
Epoch 210/350
46/46 - 0s - loss: 0.3439 - accuracy: 0.9016 - val_loss: 0.4032 - val_accuracy: 0.8983 - 259ms/epoch - 6ms/step
Epoch 211/350
46/46 - 0s - loss: 0.3387 - accuracy: 0.9000 - val_loss: 0.3702 - val_accuracy: 0.9003 - 265ms/epoch - 6ms/step
Epoch 212/350
46/46 - 0s - loss: 0.3378 - accuracy: 0.9016 - val_loss: 0.3699 - val_accuracy: 0.9080 - 281ms/epoch - 6ms/step
Epoch 213/350
46/46 - 0s - loss: 0.3760 - accuracy: 0.8876 - val_loss: 0.3798 - val_accuracy: 0.9013 - 302ms/epoch - 7ms/step
Epoch 214/350
46/46 - 0s - loss: 0.3567 - accuracy: 0.8985 - val_loss: 0.4050 - val_accuracy: 0.8977 - 288ms/epoch - 6ms/step
Epoch 215/350
46/46 - 0s - loss: 0.3738 - accuracy: 0.8888 - val_loss: 0.3724 - val_accuracy: 0.9033 - 279ms/epoch - 6ms/step
Epoch 216/350
46/46 - 0s - loss: 0.3515 - accuracy: 0.8947 - val_loss: 0.3856 - val_accuracy: 0.9037 - 279ms/epoch - 6ms/step
Epoch 217/350
46/46 - 0s - loss: 0.3556 - accuracy: 0.8994 - val_loss: 0.3873 - val_accuracy: 0.8977 - 276ms/epoch - 6ms/step
Epoch 218/350
46/46 - 0s - loss: 0.3481 - accuracy: 0.9035 - val_loss: 0.3777 - val_accuracy: 0.9000 - 276ms/epoch - 6ms/step
Epoch 219/350
46/46 - 0s - loss: 0.3438 - accuracy: 0.8994 - val_loss: 0.3655 - val_accuracy: 0.9017 - 274ms/epoch - 6ms/step
Epoch 220/350
46/46 - 0s - loss: 0.3600 - accuracy: 0.8930 - val_loss: 0.3886 - val_accuracy: 0.8990 - 275ms/epoch - 6ms/step
Epoch 221/350
46/46 - 0s - loss: 0.3421 - accuracy: 0.9049 - val_loss: 0.3601 - val_accuracy: 0.9077 - 263ms/epoch - 6ms/step
Epoch 222/350
46/46 - 0s - loss: 0.3410 - accuracy: 0.9052 - val_loss: 0.4269 - val_accuracy: 0.8900 - 267ms/epoch - 6ms/step
Epoch 223/350
46/46 - 0s - loss: 0.3515 - accuracy: 0.8972 - val_loss: 0.3626 - val_accuracy: 0.9030 - 261ms/epoch - 6ms/step
Epoch 224/350
46/46 - 0s - loss: 0.3241 - accuracy: 0.9051 - val_loss: 0.3719 - val_accuracy: 0.9033 - 262ms/epoch - 6ms/step
Epoch 225/350
46/46 - 0s - loss: 0.3269 - accuracy: 0.9106 - val_loss: 0.3627 - val_accuracy: 0.9103 - 267ms/epoch - 6ms/step
Epoch 226/350
46/46 - 0s - loss: 0.3328 - accuracy: 0.9033 - val_loss: 0.4032 - val_accuracy: 0.8977 - 262ms/epoch - 6ms/step
Epoch 227/350
46/46 - 0s - loss: 0.3397 - accuracy: 0.9023 - val_loss: 0.3810 - val_accuracy: 0.9077 - 261ms/epoch - 6ms/step
Epoch 228/350
46/46 - 0s - loss: 0.3159 - accuracy: 0.9088 - val_loss: 0.3553 - val_accuracy: 0.9113 - 258ms/epoch - 6ms/step
Epoch 229/350
46/46 - 0s - loss: 0.3325 - accuracy: 0.9078 - val_loss: 0.3678 - val_accuracy: 0.9010 - 264ms/epoch - 6ms/step
Epoch 230/350
46/46 - 0s - loss: 0.3358 - accuracy: 0.9075 - val_loss: 0.3899 - val_accuracy: 0.8980 - 257ms/epoch - 6ms/step
Epoch 231/350
46/46 - 0s - loss: 0.3176 - accuracy: 0.9072 - val_loss: 0.4139 - val_accuracy: 0.8993 - 268ms/epoch - 6ms/step
Epoch 232/350
46/46 - 0s - loss: 0.3359 - accuracy: 0.9047 - val_loss: 0.3913 - val_accuracy: 0.9030 - 260ms/epoch - 6ms/step
Epoch 233/350
46/46 - 0s - loss: 0.3417 - accuracy: 0.9044 - val_loss: 0.3848 - val_accuracy: 0.9037 - 259ms/epoch - 6ms/step
Epoch 234/350
46/46 - 0s - loss: 0.3360 - accuracy: 0.9057 - val_loss: 0.3813 - val_accuracy: 0.9003 - 261ms/epoch - 6ms/step
Epoch 235/350
46/46 - 0s - loss: 0.3346 - accuracy: 0.9030 - val_loss: 0.3618 - val_accuracy: 0.9040 - 261ms/epoch - 6ms/step
Epoch 236/350
46/46 - 0s - loss: 0.3280 - accuracy: 0.9074 - val_loss: 0.4216 - val_accuracy: 0.8887 - 262ms/epoch - 6ms/step
Epoch 237/350
46/46 - 0s - loss: 0.3229 - accuracy: 0.9065 - val_loss: 0.3657 - val_accuracy: 0.9083 - 260ms/epoch - 6ms/step
Epoch 238/350
46/46 - 0s - loss: 0.3674 - accuracy: 0.8998 - val_loss: 0.3723 - val_accuracy: 0.9083 - 273ms/epoch - 6ms/step
Epoch 239/350
46/46 - 0s - loss: 0.3313 - accuracy: 0.9016 - val_loss: 0.3841 - val_accuracy: 0.9063 - 263ms/epoch - 6ms/step
Epoch 240/350
46/46 - 0s - loss: 0.3194 - accuracy: 0.9091 - val_loss: 0.3889 - val_accuracy: 0.9010 - 270ms/epoch - 6ms/step
Epoch 241/350
46/46 - 0s - loss: 0.4101 - accuracy: 0.8839 - val_loss: 0.6152 - val_accuracy: 0.8380 - 267ms/epoch - 6ms/step
Epoch 242/350
46/46 - 0s - loss: 0.5220 - accuracy: 0.8546 - val_loss: 0.3929 - val_accuracy: 0.8980 - 263ms/epoch - 6ms/step
Epoch 243/350
46/46 - 0s - loss: 0.3372 - accuracy: 0.9012 - val_loss: 0.3458 - val_accuracy: 0.9133 - 258ms/epoch - 6ms/step
Epoch 244/350
46/46 - 0s - loss: 0.3378 - accuracy: 0.9055 - val_loss: 0.3879 - val_accuracy: 0.9040 - 266ms/epoch - 6ms/step
Epoch 245/350
46/46 - 0s - loss: 0.3195 - accuracy: 0.9086 - val_loss: 0.3414 - val_accuracy: 0.9150 - 266ms/epoch - 6ms/step
Epoch 246/350
46/46 - 0s - loss: 0.3279 - accuracy: 0.9089 - val_loss: 0.3444 - val_accuracy: 0.9117 - 257ms/epoch - 6ms/step
Epoch 247/350
46/46 - 0s - loss: 0.3039 - accuracy: 0.9144 - val_loss: 0.3367 - val_accuracy: 0.9110 - 259ms/epoch - 6ms/step
Epoch 248/350
46/46 - 0s - loss: 0.3368 - accuracy: 0.9005 - val_loss: 0.3589 - val_accuracy: 0.9060 - 259ms/epoch - 6ms/step
Epoch 249/350
46/46 - 0s - loss: 0.3056 - accuracy: 0.9154 - val_loss: 0.3740 - val_accuracy: 0.9027 - 280ms/epoch - 6ms/step
Epoch 250/350
46/46 - 0s - loss: 0.3057 - accuracy: 0.9148 - val_loss: 0.3662 - val_accuracy: 0.9073 - 275ms/epoch - 6ms/step
Epoch 251/350
46/46 - 0s - loss: 0.3145 - accuracy: 0.9118 - val_loss: 0.3566 - val_accuracy: 0.9087 - 272ms/epoch - 6ms/step
Epoch 252/350
46/46 - 0s - loss: 0.3093 - accuracy: 0.9086 - val_loss: 0.3572 - val_accuracy: 0.9090 - 267ms/epoch - 6ms/step
Epoch 253/350
46/46 - 0s - loss: 0.3183 - accuracy: 0.9060 - val_loss: 0.3503 - val_accuracy: 0.9100 - 261ms/epoch - 6ms/step
Epoch 254/350
46/46 - 0s - loss: 0.3530 - accuracy: 0.8923 - val_loss: 0.3483 - val_accuracy: 0.9097 - 258ms/epoch - 6ms/step
Epoch 255/350
46/46 - 0s - loss: 0.3117 - accuracy: 0.9121 - val_loss: 0.3854 - val_accuracy: 0.9057 - 268ms/epoch - 6ms/step
Epoch 256/350
46/46 - 0s - loss: 0.3757 - accuracy: 0.8947 - val_loss: 0.3887 - val_accuracy: 0.8977 - 265ms/epoch - 6ms/step
Epoch 257/350
46/46 - 0s - loss: 0.3284 - accuracy: 0.9044 - val_loss: 0.3682 - val_accuracy: 0.9040 - 259ms/epoch - 6ms/step
Epoch 258/350
46/46 - 0s - loss: 0.3222 - accuracy: 0.9099 - val_loss: 0.3403 - val_accuracy: 0.9060 - 265ms/epoch - 6ms/step
Epoch 259/350
46/46 - 0s - loss: 0.3464 - accuracy: 0.9027 - val_loss: 0.3900 - val_accuracy: 0.9000 - 275ms/epoch - 6ms/step
Epoch 260/350
46/46 - 0s - loss: 0.3659 - accuracy: 0.8980 - val_loss: 0.3624 - val_accuracy: 0.9030 - 271ms/epoch - 6ms/step
Epoch 261/350
46/46 - 0s - loss: 0.3182 - accuracy: 0.9085 - val_loss: 0.3900 - val_accuracy: 0.9030 - 265ms/epoch - 6ms/step
Epoch 262/350
46/46 - 0s - loss: 0.3093 - accuracy: 0.9112 - val_loss: 0.3533 - val_accuracy: 0.9120 - 258ms/epoch - 6ms/step
Epoch 263/350
46/46 - 0s - loss: 0.3268 - accuracy: 0.9083 - val_loss: 0.3799 - val_accuracy: 0.9050 - 265ms/epoch - 6ms/step
Epoch 264/350
46/46 - 0s - loss: 0.3139 - accuracy: 0.9153 - val_loss: 0.3485 - val_accuracy: 0.9113 - 262ms/epoch - 6ms/step
Epoch 265/350
46/46 - 0s - loss: 0.2892 - accuracy: 0.9174 - val_loss: 0.3625 - val_accuracy: 0.9063 - 262ms/epoch - 6ms/step
Epoch 266/350
46/46 - 0s - loss: 0.3306 - accuracy: 0.9055 - val_loss: 0.3527 - val_accuracy: 0.9140 - 260ms/epoch - 6ms/step
Epoch 267/350
46/46 - 0s - loss: 0.2809 - accuracy: 0.9195 - val_loss: 0.3587 - val_accuracy: 0.9140 - 262ms/epoch - 6ms/step
Epoch 268/350
46/46 - 0s - loss: 0.2985 - accuracy: 0.9092 - val_loss: 0.3297 - val_accuracy: 0.9150 - 261ms/epoch - 6ms/step
Epoch 269/350
46/46 - 0s - loss: 0.3032 - accuracy: 0.9148 - val_loss: 0.3547 - val_accuracy: 0.9103 - 263ms/epoch - 6ms/step
Epoch 270/350
46/46 - 0s - loss: 0.3013 - accuracy: 0.9089 - val_loss: 0.3385 - val_accuracy: 0.9120 - 268ms/epoch - 6ms/step
Epoch 271/350
46/46 - 0s - loss: 0.2946 - accuracy: 0.9155 - val_loss: 0.3516 - val_accuracy: 0.9133 - 267ms/epoch - 6ms/step
Epoch 272/350
46/46 - 0s - loss: 0.3532 - accuracy: 0.9029 - val_loss: 0.3933 - val_accuracy: 0.8857 - 265ms/epoch - 6ms/step
Epoch 273/350
46/46 - 0s - loss: 0.3326 - accuracy: 0.9036 - val_loss: 0.3711 - val_accuracy: 0.9053 - 270ms/epoch - 6ms/step
Epoch 274/350
46/46 - 0s - loss: 0.3135 - accuracy: 0.9097 - val_loss: 0.3424 - val_accuracy: 0.9163 - 264ms/epoch - 6ms/step
Epoch 275/350
46/46 - 0s - loss: 0.3006 - accuracy: 0.9174 - val_loss: 0.3435 - val_accuracy: 0.9157 - 267ms/epoch - 6ms/step
Epoch 276/350
46/46 - 0s - loss: 0.3048 - accuracy: 0.9193 - val_loss: 0.3509 - val_accuracy: 0.9143 - 276ms/epoch - 6ms/step
Epoch 277/350
46/46 - 0s - loss: 0.2998 - accuracy: 0.9145 - val_loss: 0.3431 - val_accuracy: 0.9130 - 286ms/epoch - 6ms/step
Epoch 278/350
46/46 - 0s - loss: 0.3434 - accuracy: 0.9049 - val_loss: 0.3591 - val_accuracy: 0.9013 - 279ms/epoch - 6ms/step
Epoch 279/350
46/46 - 0s - loss: 0.3039 - accuracy: 0.9142 - val_loss: 0.3287 - val_accuracy: 0.9190 - 273ms/epoch - 6ms/step
Epoch 280/350
46/46 - 0s - loss: 0.3178 - accuracy: 0.9089 - val_loss: 0.3592 - val_accuracy: 0.9063 - 271ms/epoch - 6ms/step
Epoch 281/350
46/46 - 0s - loss: 0.3170 - accuracy: 0.9097 - val_loss: 0.4152 - val_accuracy: 0.8907 - 283ms/epoch - 6ms/step
Epoch 282/350
46/46 - 0s - loss: 0.3212 - accuracy: 0.9045 - val_loss: 0.3838 - val_accuracy: 0.8987 - 274ms/epoch - 6ms/step
Epoch 283/350
46/46 - 0s - loss: 0.3076 - accuracy: 0.9133 - val_loss: 0.3732 - val_accuracy: 0.9073 - 276ms/epoch - 6ms/step
Epoch 284/350
46/46 - 0s - loss: 0.3003 - accuracy: 0.9149 - val_loss: 0.3534 - val_accuracy: 0.9093 - 283ms/epoch - 6ms/step
Epoch 285/350
46/46 - 0s - loss: 0.3165 - accuracy: 0.9122 - val_loss: 0.3721 - val_accuracy: 0.9073 - 277ms/epoch - 6ms/step
Epoch 286/350
46/46 - 0s - loss: 0.3258 - accuracy: 0.9085 - val_loss: 0.3428 - val_accuracy: 0.9057 - 283ms/epoch - 6ms/step
Epoch 287/350
46/46 - 0s - loss: 0.3177 - accuracy: 0.9103 - val_loss: 0.3427 - val_accuracy: 0.9137 - 269ms/epoch - 6ms/step
Epoch 288/350
46/46 - 0s - loss: 0.3092 - accuracy: 0.9103 - val_loss: 0.3497 - val_accuracy: 0.9087 - 266ms/epoch - 6ms/step
Epoch 289/350
46/46 - 0s - loss: 0.2852 - accuracy: 0.9173 - val_loss: 0.3583 - val_accuracy: 0.9117 - 264ms/epoch - 6ms/step
Epoch 290/350
46/46 - 0s - loss: 0.3169 - accuracy: 0.9111 - val_loss: 0.4464 - val_accuracy: 0.8927 - 260ms/epoch - 6ms/step
Epoch 291/350
46/46 - 0s - loss: 0.3706 - accuracy: 0.8947 - val_loss: 0.4179 - val_accuracy: 0.8903 - 261ms/epoch - 6ms/step
Epoch 292/350
46/46 - 0s - loss: 0.2940 - accuracy: 0.9143 - val_loss: 0.3323 - val_accuracy: 0.9183 - 261ms/epoch - 6ms/step
Epoch 293/350
46/46 - 0s - loss: 0.2941 - accuracy: 0.9163 - val_loss: 0.3728 - val_accuracy: 0.9063 - 281ms/epoch - 6ms/step
Epoch 294/350
46/46 - 0s - loss: 0.3057 - accuracy: 0.9139 - val_loss: 0.3523 - val_accuracy: 0.9100 - 269ms/epoch - 6ms/step
Epoch 295/350
46/46 - 0s - loss: 0.2778 - accuracy: 0.9215 - val_loss: 0.3296 - val_accuracy: 0.9190 - 261ms/epoch - 6ms/step
Epoch 296/350
46/46 - 0s - loss: 0.2933 - accuracy: 0.9154 - val_loss: 0.3755 - val_accuracy: 0.9107 - 273ms/epoch - 6ms/step
Epoch 297/350
46/46 - 0s - loss: 0.3672 - accuracy: 0.9013 - val_loss: 0.3507 - val_accuracy: 0.9107 - 276ms/epoch - 6ms/step
Epoch 298/350
46/46 - 0s - loss: 0.3148 - accuracy: 0.9093 - val_loss: 0.3529 - val_accuracy: 0.9117 - 272ms/epoch - 6ms/step
Epoch 299/350
46/46 - 0s - loss: 0.3007 - accuracy: 0.9134 - val_loss: 0.3620 - val_accuracy: 0.9037 - 271ms/epoch - 6ms/step
Epoch 300/350
46/46 - 0s - loss: 0.3003 - accuracy: 0.9140 - val_loss: 0.3841 - val_accuracy: 0.9057 - 262ms/epoch - 6ms/step
Epoch 301/350
46/46 - 0s - loss: 0.2893 - accuracy: 0.9233 - val_loss: 0.3465 - val_accuracy: 0.9147 - 260ms/epoch - 6ms/step
Epoch 302/350
46/46 - 0s - loss: 0.2935 - accuracy: 0.9171 - val_loss: 0.3743 - val_accuracy: 0.8987 - 262ms/epoch - 6ms/step
Epoch 303/350
46/46 - 0s - loss: 0.3661 - accuracy: 0.8961 - val_loss: 0.3390 - val_accuracy: 0.9127 - 266ms/epoch - 6ms/step
Epoch 304/350
46/46 - 0s - loss: 0.3069 - accuracy: 0.9161 - val_loss: 0.3402 - val_accuracy: 0.9090 - 261ms/epoch - 6ms/step
Epoch 305/350
46/46 - 0s - loss: 0.3099 - accuracy: 0.9127 - val_loss: 0.3410 - val_accuracy: 0.9147 - 267ms/epoch - 6ms/step
Epoch 306/350
46/46 - 0s - loss: 0.3101 - accuracy: 0.9139 - val_loss: 0.3579 - val_accuracy: 0.9090 - 272ms/epoch - 6ms/step
Epoch 307/350
46/46 - 0s - loss: 0.3227 - accuracy: 0.9098 - val_loss: 0.3440 - val_accuracy: 0.9113 - 270ms/epoch - 6ms/step
Epoch 308/350
46/46 - 0s - loss: 0.2812 - accuracy: 0.9188 - val_loss: 0.3942 - val_accuracy: 0.9020 - 267ms/epoch - 6ms/step
Epoch 309/350
46/46 - 0s - loss: 0.2740 - accuracy: 0.9237 - val_loss: 0.3571 - val_accuracy: 0.9117 - 270ms/epoch - 6ms/step
Epoch 310/350
46/46 - 0s - loss: 0.2968 - accuracy: 0.9181 - val_loss: 0.3553 - val_accuracy: 0.9127 - 269ms/epoch - 6ms/step
Epoch 311/350
46/46 - 0s - loss: 0.2913 - accuracy: 0.9187 - val_loss: 0.3420 - val_accuracy: 0.9123 - 268ms/epoch - 6ms/step
Epoch 312/350
46/46 - 0s - loss: 0.2984 - accuracy: 0.9167 - val_loss: 0.3184 - val_accuracy: 0.9153 - 270ms/epoch - 6ms/step
Epoch 313/350
46/46 - 0s - loss: 0.2954 - accuracy: 0.9174 - val_loss: 0.3269 - val_accuracy: 0.9157 - 275ms/epoch - 6ms/step
Epoch 314/350
46/46 - 0s - loss: 0.2934 - accuracy: 0.9184 - val_loss: 0.3558 - val_accuracy: 0.9103 - 268ms/epoch - 6ms/step
Epoch 315/350
46/46 - 0s - loss: 0.2781 - accuracy: 0.9189 - val_loss: 0.3498 - val_accuracy: 0.9080 - 262ms/epoch - 6ms/step
Epoch 316/350
46/46 - 0s - loss: 0.2737 - accuracy: 0.9208 - val_loss: 0.3501 - val_accuracy: 0.9137 - 263ms/epoch - 6ms/step
Epoch 317/350
46/46 - 0s - loss: 0.2602 - accuracy: 0.9270 - val_loss: 0.3427 - val_accuracy: 0.9137 - 262ms/epoch - 6ms/step
Epoch 318/350
46/46 - 0s - loss: 0.2839 - accuracy: 0.9201 - val_loss: 0.3428 - val_accuracy: 0.9130 - 271ms/epoch - 6ms/step
Epoch 319/350
46/46 - 0s - loss: 0.2905 - accuracy: 0.9197 - val_loss: 0.3565 - val_accuracy: 0.9073 - 274ms/epoch - 6ms/step
Epoch 320/350
46/46 - 0s - loss: 0.3004 - accuracy: 0.9163 - val_loss: 0.3413 - val_accuracy: 0.9113 - 259ms/epoch - 6ms/step
Epoch 321/350
46/46 - 0s - loss: 0.3201 - accuracy: 0.9134 - val_loss: 0.3710 - val_accuracy: 0.9040 - 262ms/epoch - 6ms/step
Epoch 322/350
46/46 - 0s - loss: 0.3000 - accuracy: 0.9186 - val_loss: 0.3358 - val_accuracy: 0.9167 - 264ms/epoch - 6ms/step
Epoch 323/350
46/46 - 0s - loss: 0.2746 - accuracy: 0.9229 - val_loss: 0.3237 - val_accuracy: 0.9170 - 264ms/epoch - 6ms/step
Epoch 324/350
46/46 - 0s - loss: 0.2827 - accuracy: 0.9224 - val_loss: 0.3358 - val_accuracy: 0.9123 - 260ms/epoch - 6ms/step
Epoch 325/350
46/46 - 0s - loss: 0.2788 - accuracy: 0.9212 - val_loss: 0.3734 - val_accuracy: 0.9020 - 267ms/epoch - 6ms/step
Epoch 326/350
46/46 - 0s - loss: 0.2739 - accuracy: 0.9232 - val_loss: 0.3570 - val_accuracy: 0.9110 - 270ms/epoch - 6ms/step
Epoch 327/350
46/46 - 0s - loss: 0.2917 - accuracy: 0.9166 - val_loss: 0.3361 - val_accuracy: 0.9157 - 269ms/epoch - 6ms/step
Epoch 328/350
46/46 - 0s - loss: 0.2548 - accuracy: 0.9281 - val_loss: 0.3346 - val_accuracy: 0.9187 - 259ms/epoch - 6ms/step
Epoch 329/350
46/46 - 0s - loss: 0.2985 - accuracy: 0.9161 - val_loss: 0.4006 - val_accuracy: 0.9007 - 264ms/epoch - 6ms/step
Epoch 330/350
46/46 - 0s - loss: 0.3046 - accuracy: 0.9148 - val_loss: 0.3294 - val_accuracy: 0.9167 - 257ms/epoch - 6ms/step
Epoch 331/350
46/46 - 0s - loss: 0.2618 - accuracy: 0.9270 - val_loss: 0.3667 - val_accuracy: 0.9090 - 262ms/epoch - 6ms/step
Epoch 332/350
46/46 - 0s - loss: 0.3251 - accuracy: 0.9070 - val_loss: 0.3311 - val_accuracy: 0.9150 - 272ms/epoch - 6ms/step
Epoch 333/350
46/46 - 0s - loss: 0.2826 - accuracy: 0.9216 - val_loss: 0.3546 - val_accuracy: 0.9103 - 280ms/epoch - 6ms/step
Epoch 334/350
46/46 - 0s - loss: 0.2651 - accuracy: 0.9227 - val_loss: 0.3478 - val_accuracy: 0.9140 - 274ms/epoch - 6ms/step
Epoch 335/350
46/46 - 0s - loss: 0.2837 - accuracy: 0.9177 - val_loss: 0.3315 - val_accuracy: 0.9153 - 277ms/epoch - 6ms/step
Epoch 336/350
46/46 - 0s - loss: 0.2878 - accuracy: 0.9216 - val_loss: 0.3393 - val_accuracy: 0.9143 - 275ms/epoch - 6ms/step
Epoch 337/350
46/46 - 0s - loss: 0.2842 - accuracy: 0.9237 - val_loss: 0.3657 - val_accuracy: 0.9087 - 264ms/epoch - 6ms/step
Epoch 338/350
46/46 - 0s - loss: 0.2933 - accuracy: 0.9200 - val_loss: 0.3487 - val_accuracy: 0.9223 - 262ms/epoch - 6ms/step
Epoch 339/350
46/46 - 0s - loss: 0.2846 - accuracy: 0.9176 - val_loss: 0.3697 - val_accuracy: 0.9113 - 263ms/epoch - 6ms/step
Epoch 340/350
46/46 - 0s - loss: 0.2701 - accuracy: 0.9246 - val_loss: 0.3477 - val_accuracy: 0.9143 - 266ms/epoch - 6ms/step
Epoch 341/350
46/46 - 0s - loss: 0.2785 - accuracy: 0.9227 - val_loss: 0.3757 - val_accuracy: 0.9103 - 263ms/epoch - 6ms/step
Epoch 342/350
46/46 - 0s - loss: 0.2703 - accuracy: 0.9252 - val_loss: 0.3618 - val_accuracy: 0.9033 - 270ms/epoch - 6ms/step
Epoch 343/350
46/46 - 0s - loss: 0.2968 - accuracy: 0.9206 - val_loss: 0.3711 - val_accuracy: 0.9117 - 263ms/epoch - 6ms/step
Epoch 344/350
46/46 - 0s - loss: 0.2733 - accuracy: 0.9260 - val_loss: 0.3349 - val_accuracy: 0.9160 - 262ms/epoch - 6ms/step
Epoch 345/350
46/46 - 0s - loss: 0.2723 - accuracy: 0.9246 - val_loss: 0.3298 - val_accuracy: 0.9210 - 272ms/epoch - 6ms/step
Epoch 346/350
46/46 - 0s - loss: 0.2813 - accuracy: 0.9229 - val_loss: 0.3485 - val_accuracy: 0.9137 - 267ms/epoch - 6ms/step
Epoch 347/350
46/46 - 0s - loss: 0.2804 - accuracy: 0.9228 - val_loss: 0.3561 - val_accuracy: 0.9090 - 264ms/epoch - 6ms/step
Epoch 348/350
46/46 - 0s - loss: 0.3117 - accuracy: 0.9117 - val_loss: 0.3435 - val_accuracy: 0.9183 - 266ms/epoch - 6ms/step
Epoch 349/350
46/46 - 0s - loss: 0.3076 - accuracy: 0.9163 - val_loss: 0.3280 - val_accuracy: 0.9170 - 276ms/epoch - 6ms/step
Epoch 350/350
46/46 - 0s - loss: 0.2660 - accuracy: 0.9246 - val_loss: 0.4120 - val_accuracy: 0.9050 - 266ms/epoch - 6ms/step
94/94 - 0s - loss: 0.4120 - accuracy: 0.9050 - 370ms/epoch - 4ms/step
Baseline Error: 9.50%

We can see that the model is being fitted very well with a high accuracy (over 90%) as well as there not being much overfitting. This a massive improvement from the original model, which can be seen below

model_31new.save('Goat31Model.h5') #save the model to evaluate on test data later on

df31

	Model Name	Train Accuracy	Validation Accuracy	Train Loss	Validation Loss	History
0	Base 31 Model	0.780128	0.537667	0.679253	1.505984	<keras.callbacks.History object at 0x7f5a1785d...
1	LeNet	0.728622	0.576000	0.878840	1.332178	<keras.callbacks.History object at 0x7f5a0c5dd...
2	AlexNet	0.105782	0.066667	2.637130	2.785964	<keras.callbacks.History object at 0x7f5a0c3bb...
3	RegularizedModel	0.963779	0.805333	0.181553	1.063044	<keras.callbacks.History object at 0x7f59ec339...
4	RegularizedDOModel	0.870403	0.790000	0.413026	1.056112	<keras.callbacks.History object at 0x7f59ec0b8...
5	RegularizedDOModel2	0.946500	0.799000	0.193856	1.499866	<keras.callbacks.History object at 0x7f598a934...
6	LearningRateDecayModel	0.797630	0.763333	0.573027	0.902065	<keras.callbacks.History object at 0x7f5960531...
7	ExtraRegularReducedModel	0.868520	0.891333	0.445079	0.408669	<keras.callbacks.History object at 0x7f59602f6...
8	NormalizedModel	0.924568	0.905000	0.265987	0.412032	<keras.callbacks.History object at 0x7f5971bb2...

plt.figure(figsize=(15, 8))

for index, row in df31[['Model Name', 'History']].iterrows():
    model_name = row['Model Name']
    history = row['History']
    sns.lineplot(x=range(1, len(history.history['val_accuracy']) + 1), y=history.history['val_accuracy'], label=f'{model_name} Validation')

plt.title('31x31 Model Accuracies Over Epochs')
plt.xlabel('Epochs')
plt.ylabel('Accuracy')
plt.legend()

plt.show() 

plt.figure(figsize=(15, 8))

for index, row in df31[['Model Name', 'History']].iterrows():
    model_name = row['Model Name']
    history = row['History']

    accuracy_diff = [train_acc - val_acc for train_acc, val_acc in zip(history.history['accuracy'], history.history['val_accuracy'])]

    sns.lineplot(x=range(1, len(accuracy_diff) + 1), y=accuracy_diff, label=f'{model_name}')

plt.title('Training Accuracy - Validation Accuracy Difference Over Epochs(31x31)')
plt.xlabel('Epochs')
plt.ylabel('Accuracy Difference (To see overfitting)')

plt.legend()

plt.show()

Graphs Evaluation

As we can see, the model with the least overfitting is the model with increased dropout layers as well as the model with batch normalization. If we are looking at which model has the most consistent performance, it can be said that the model with batch normalization is the best, from the val accuracy graph, we can also see that the model with batch normalization has the highest val accuracy and converges faster. However before we make our final judgement we can test both models on the test set and see which model performs better.

#convert tets to 31x31 
X_test_resized = tf.image.resize(X_test, (31, 31))
X_test31 = X_test_resized.numpy().reshape(X_test_resized.shape[0], 31, 31, 1).astype('float32') / 255
test_labels = to_categorical(y_test)

#load the Model with Batch Normalization
model31_BN = models.load_model('Goat31Model.h5')


#evaluate the model on the test set
score = model31_BN.evaluate(X_test31, test_labels, verbose=0)
print("Test loss:", score[0])
print("Test accuracy:", score[1])

Test loss: 0.3477824330329895
Test accuracy: 0.9086666703224182

#load the other model for comparison 


model31_reg = models.load_model('extraregularized31.h5')


#evaluate the model on the test set
score = model31_reg.evaluate(X_test31, test_labels, verbose=0)
print("Test loss:", score[0])
print("Test accuracy:", score[1])

Test loss: 0.39638611674308777
Test accuracy: 0.9003333449363708

Model: Regularized Model

• Test Loss: 0.3756
• Test Accuracy: 89.99%

Model: Batch Normalized Model

• Test Loss: 0.4186
• Test Accuracy: 89.43%

Comparison:

• Accuracy: Regularized Model has higher test accuracy compared to Batch Normalized Model.

• Loss: Regularized Model has a lower test loss compared to Batch Normalized Model.

Conclusion:

The Regularized Model performs better on the test set, having higher accuracy and lower loss than the Batch Normalized Model. Therefore, the Regularized Model is the preferred model between the two.

#Compare the Best model to our base model

base31model = models.load_model('Base31Model.h5')

score = base31model.evaluate(X_test31, test_labels, verbose=0)
print("Test loss:", score[0])
print("Test accuracy:", score[1])

Test loss: 1.5366089344024658
Test accuracy: 0.5286666750907898

31x31 Super Model with feature eng and data augment

As part of the assignment requirements, we are needed to do feature engineering and/or data augmentation. We can use data augmentation to balance the imbalanced data sets instead of setting class_weights. We can then train a model using this new "balanced data" and compare the results of setting class_weights vs using data augmentation.

Data Augmentation

What is Data Augmentation?

Data augmentation is a technique widely used in machine learning and computer vision to artificially increase the diversity of a training dataset by applying various transformations to the existing data. The goal is to create new variations of the original images while preserving their semantic content. This approach is particularly useful in image classification tasks.

How Does Data Augmentation Work?

Data augmentation involves applying a set of image transformations to the training data, effectively creating new instances that are variations of the original samples. Common transformations include:

• Rotation: Rotating the image by a certain angle.
• Flip: Flipping the image horizontally or vertically.
• Zoom: Zooming in or out of the image.
• Translation: Shifting the image horizontally or vertically.
• Brightness and Contrast Adjustment: Changing the brightness and contrast levels.

By introducing these variations, the model becomes more robust and generalizes better to unseen data. Data augmentation helps prevent overfitting by exposing the model to a more diverse range of input patterns during training.

Why Use Data Augmentation?

1. Increased Dataset Size: Data augmentation effectively expands the training dataset, providing the model with more examples to learn from.
2. Improved Generalization: The augmented data exposes the model to a broader range of scenarios, improving its ability to generalize to unseen variations in the test set.
3. Reduced Overfitting: By presenting the model with diverse examples, data augmentation helps prevent overfitting and enhances the model's ability to handle real-world variations.

def augmentor(x_batch):
    datagen = ImageDataGenerator(
        rotation_range=40,
        width_shift_range=0.2,
        height_shift_range=0.2,
        shear_range=0.2,
        zoom_range=0.2,
        horizontal_flip=True,
        fill_mode='nearest'
    )
    # Specify the batch_size to match the size of x_batch
    batch_size = len(x_batch)
    x_batch_augmented = datagen.flow(x_batch, shuffle=False, batch_size=batch_size).next()

    return x_batch_augmented

# Test the function on our 128x128 images
x_batch = X_train31[:3]
print(x_batch.shape)
x_batch_augmented = augmentor(x_batch)

# Print the original and augmented images
fig, axes = plt.subplots(2, 3, figsize=(15, 8))

# Original Images
for i in range(3):
    ax = axes[0, i]
    ax.imshow(x_batch[i].reshape(31, 31), cmap='gray')
    ax.axis('off')

# Augmented Images
for i in range(3):
    ax = axes[1, i]
    ax.imshow(x_batch_augmented[i].reshape(31,31), cmap='gray')
    ax.axis('off')

fig.suptitle('Original (Top) vs Augmented (Bottom) Images', fontsize=16)
plt.show()

(3, 31, 31, 1)

Distribution of data

We recall the previous distribution of classes from the bar graph that we plotted:

We can use data augmentation to balance out the number of data entries

class_counts

# this is on the old training data

{'Cauliflower': 587,
 'Cucumber': 812,
 'Broccoli': 750,
 'Bean': 780,
 'Pumpkin': 814,
 'Tomato': 955,
 'Papaya': 566,
 'Bottle_Gourd': 441,
 'Cabbage': 503,
 'Brinjal': 868,
 'Capsicum': 351,
 'Potato': 377,
 'Radish': 248,
 'Bitter_Gourd': 720,
 'Carrot': 256}

Feature engineered images

As a form of feature engineering, i have removed the hands from the images in the training data. Hopefully this will make the data more "vegetable like" allowing the models to learn and differentiate vegetables better.

train_data = tf.keras.utils.image_dataset_from_directory(
    'train_aug/',
    labels='inferred',
    label_mode='int',
    class_names=None,
    color_mode='grayscale',
    batch_size = 200,
    image_size=(224, 224),
    shuffle=True,
    seed=42,
    validation_split=None,
    subset=None,
    interpolation='bilinear',
    crop_to_aspect_ratio=False,
)

Found 8807 files belonging to 15 classes.

Balance class

We can see that the class with the highest number of data entries is Tomato with 955 entries, let us generate a new dictionary with the amount of entries we need to "generate" in order to balance out the dataset.

#change the labels to be in int format
class_counts = {}

for images, labels in train_data:
    for label in labels.numpy():
        if label in class_counts:
            class_counts[label] += 1
        else:
            class_counts[label] = 1

class_counts

2023-11-26 13:25:44.579894: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'Placeholder/_4' with dtype int32 and shape [8807]
	 [[{{node Placeholder/_4}}]]
2023-11-26 13:25:44.580457: I tensorflow/core/common_runtime/executor.cc:1197] [/device:CPU:0] (DEBUG INFO) Executor start aborting (this does not indicate an error and you can ignore this message): INVALID_ARGUMENT: You must feed a value for placeholder tensor 'Placeholder/_0' with dtype string and shape [8807]
	 [[{{node Placeholder/_0}}]]

{14: 955,
 2: 420,
 0: 741,
 4: 736,
 1: 701,
 9: 750,
 10: 563,
 12: 811,
 6: 346,
 11: 373,
 5: 501,
 13: 247,
 3: 826,
 8: 581,
 7: 256}

augmentation_dict = {class_name: 955 - class_count for class_name, class_count in class_counts.items()}

Feature Engineered data

Compared to the original train data with hands, our engineered data has less data, however we can say that the quality of the data has gone up.

# Convert datasets to NumPy arrays
train_images, train_labels_augmented = zip(*[(image, label) for image, label in train_data.as_numpy_iterator()])

#train set
X_train_aug = np.concatenate(train_images, axis=0)
y_train = np.concatenate(train_labels_augmented, axis=0)

# Resize images to (31, 31) using TensorFlow
X_train_resized = tf.image.resize(X_train_aug, (31, 31))

# Reshape and normalize
X_train31_aug = X_train_resized.numpy().reshape(X_train_resized.shape[0], 31, 31, 1).astype('float32') / 255

def augmentfill(y_train, augmentation_dict, X_train_aug, augmentor=augmentor ):
    augmented_data_list = []
    for category_index in np.unique(y_train):
        category_mask = (y_train == category_index)
        selected_images = X_train_aug[category_mask]

        # Ensure the category_index exists in augmentation_dict
        if category_index in augmentation_dict:

            # Check if there are samples in selected_images
            if len(selected_images) > 0:
                target_augmentation_count = augmentation_dict[category_index]
                current_count = 0
                while current_count < target_augmentation_count:
                    # Calculate how many images to take in this iteration
                    images_to_take = min(target_augmentation_count - current_count, len(selected_images))
                    print(images_to_take)
                    x_batch = selected_images[:images_to_take]
                    if len(x_batch) > 0:
                        x_batch_augmented = augmentor(x_batch)
                        augmented_data_list.append((x_batch_augmented, [category_index] * images_to_take))
                        current_count += images_to_take
                        print(f'Augmented {current_count} images out of {target_augmentation_count}')
                    else:
                        print(f'No augmentation needed to fill for {category_index}')
                        break
                print(f'Category {category_index}: Augmented {current_count} images out of {target_augmentation_count}')
            else:
                print(f"No samples for category {category_index}")
        else:
            print(f"No augmentation specified for category {category_index}")

    # Convert the augmented data list to NumPy arrays
    augmented_data_images, augmented_data_labels = zip(*augmented_data_list)

    # Convert augmented_data_labels to a flattened 1D array
    flattened_augmented_labels = np.concatenate(augmented_data_labels, axis=0)

    # Concatenate augmented data with the original data
    X_train31_augmented = np.concatenate([X_train_aug] + list(augmented_data_images), axis=0)
    y_train_augmented = np.concatenate([y_train] + [flattened_augmented_labels], axis=0)

    return X_train31_augmented, y_train_augmented

X_train31_augmented, y_train_augmented = augmentfill(y_train, augmentation_dict, X_train31_aug)

214
Augmented 214 images out of 214
Category 0: Augmented 214 images out of 214
254
Augmented 254 images out of 254
Category 1: Augmented 254 images out of 254
420
Augmented 420 images out of 535
115
Augmented 535 images out of 535
Category 2: Augmented 535 images out of 535
129
Augmented 129 images out of 129
Category 3: Augmented 129 images out of 129
219
Augmented 219 images out of 219
Category 4: Augmented 219 images out of 219
454
Augmented 454 images out of 454
Category 5: Augmented 454 images out of 454
346
Augmented 346 images out of 609
263
Augmented 609 images out of 609
Category 6: Augmented 609 images out of 609
256
Augmented 256 images out of 699
256
Augmented 512 images out of 699
187
Augmented 699 images out of 699
Category 7: Augmented 699 images out of 699
374
Augmented 374 images out of 374
Category 8: Augmented 374 images out of 374
205
Augmented 205 images out of 205
Category 9: Augmented 205 images out of 205
392
Augmented 392 images out of 392
Category 10: Augmented 392 images out of 392
373
Augmented 373 images out of 582
209
Augmented 582 images out of 582
Category 11: Augmented 582 images out of 582
144
Augmented 144 images out of 144
Category 12: Augmented 144 images out of 144
247
Augmented 247 images out of 708
247
Augmented 494 images out of 708
214
Augmented 708 images out of 708
Category 13: Augmented 708 images out of 708
Category 14: Augmented 0 images out of 0

values, counts = np.unique(y_train_augmented, return_counts=True)
print(values)
print(counts)

[ 0  1  2  3  4  5  6  7  8  9 10 11 12 13 14]
[955 955 955 955 955 955 955 955 955 955 955 955 955 955 955]

train_labels_augmented = to_categorical(y_train_augmented)

We can now start with the modeling, we will be trying variations of our top models on

• just augmented data
• only weights (technically we already have this)

def GoatModel(): #not to be confused with Goat31model, this is the model without batch norm
    model_31do = models.Sequential([
        layers.Conv2D(32, (3, 3), activation='relu', input_shape=(31, 31, 1)),
        layers.MaxPooling2D((2, 2)),
        layers.Dropout(0.2),

        layers.Conv2D(64, (3, 3), activation='relu'),
        layers.MaxPooling2D((2, 2)),
        layers.Dropout(0.2),

        layers.Conv2D(64, (3, 3), activation='relu'),
        layers.MaxPooling2D((2, 2)),
        layers.Dropout(0.2),

        layers.Flatten(),

        layers.Dense(64, activation='relu', kernel_regularizer=regularizers.l2(0.001)),
        layers.Dropout(0.5),

        layers.Dense(15, activation='softmax')
    ])
    return model_31do

df31augmentationcompare = pd.DataFrame(columns=['Model Name', 'Train Accuracy', 'test Accuracy', 'Val Loss', 'Val Loss', 'History'])
#with just weights
model = GoatModel()

model.compile(optimizer='adam',
                loss= 'categorical_crossentropy',
                metrics=['accuracy'])

history = model.fit(X_train31, train_labels, validation_data=(X_val31, validation_labels),
                    epochs=300, batch_size=200, verbose=2, class_weight = class_weight)

scores = model.evaluate(X_val31, validation_labels, verbose=2)
print("Baseline Error: %.2f%%" % (100-scores[1]*100))
plot_history(history)
df31augmentationcompare.loc[len(df31augmentationcompare)] = ['With Class Weights', history.history['accuracy'][-1], history.history['val_accuracy'][-1], history.history['loss'][-1], history.history['val_loss'][-1], history]

Epoch 1/300

2023-11-26 13:25:51.725383: E tensorflow/core/grappler/optimizers/meta_optimizer.cc:954] layout failed: INVALID_ARGUMENT: Size of values 0 does not match size of permutation 4 @ fanin shape insequential_12/dropout_11/dropout/SelectV2-2-TransposeNHWCToNCHW-LayoutOptimizer

46/46 - 2s - loss: 3.2633 - accuracy: 0.1007 - val_loss: 2.8381 - val_accuracy: 0.0667 - 2s/epoch - 52ms/step
Epoch 2/300
46/46 - 0s - loss: 3.1237 - accuracy: 0.1120 - val_loss: 2.7994 - val_accuracy: 0.0943 - 260ms/epoch - 6ms/step
Epoch 3/300
46/46 - 0s - loss: 2.9289 - accuracy: 0.1595 - val_loss: 2.6474 - val_accuracy: 0.1303 - 253ms/epoch - 6ms/step
Epoch 4/300
46/46 - 0s - loss: 2.7005 - accuracy: 0.2127 - val_loss: 2.4420 - val_accuracy: 0.1947 - 247ms/epoch - 5ms/step
Epoch 5/300
46/46 - 0s - loss: 2.5595 - accuracy: 0.2463 - val_loss: 2.2509 - val_accuracy: 0.2713 - 248ms/epoch - 5ms/step
Epoch 6/300
46/46 - 0s - loss: 2.3929 - accuracy: 0.2812 - val_loss: 2.1818 - val_accuracy: 0.2890 - 251ms/epoch - 5ms/step
Epoch 7/300
46/46 - 0s - loss: 2.3074 - accuracy: 0.3107 - val_loss: 2.1674 - val_accuracy: 0.3053 - 248ms/epoch - 5ms/step
Epoch 8/300
46/46 - 0s - loss: 2.2328 - accuracy: 0.3238 - val_loss: 2.0238 - val_accuracy: 0.3557 - 250ms/epoch - 5ms/step
Epoch 9/300
46/46 - 0s - loss: 2.1702 - accuracy: 0.3480 - val_loss: 1.9389 - val_accuracy: 0.3893 - 246ms/epoch - 5ms/step
Epoch 10/300
46/46 - 0s - loss: 2.0804 - accuracy: 0.3664 - val_loss: 1.9264 - val_accuracy: 0.3757 - 254ms/epoch - 6ms/step
Epoch 11/300
46/46 - 0s - loss: 2.0221 - accuracy: 0.3848 - val_loss: 1.8172 - val_accuracy: 0.4137 - 243ms/epoch - 5ms/step
Epoch 12/300
46/46 - 0s - loss: 1.9702 - accuracy: 0.3941 - val_loss: 1.7433 - val_accuracy: 0.4340 - 240ms/epoch - 5ms/step
Epoch 13/300
46/46 - 0s - loss: 1.9429 - accuracy: 0.3980 - val_loss: 1.7278 - val_accuracy: 0.4370 - 244ms/epoch - 5ms/step
Epoch 14/300
46/46 - 0s - loss: 1.8685 - accuracy: 0.4206 - val_loss: 1.7402 - val_accuracy: 0.4223 - 241ms/epoch - 5ms/step
Epoch 15/300
46/46 - 0s - loss: 1.8718 - accuracy: 0.4221 - val_loss: 1.6568 - val_accuracy: 0.4587 - 240ms/epoch - 5ms/step
Epoch 16/300
46/46 - 0s - loss: 1.8170 - accuracy: 0.4366 - val_loss: 1.5847 - val_accuracy: 0.4910 - 243ms/epoch - 5ms/step
Epoch 17/300
46/46 - 0s - loss: 1.7809 - accuracy: 0.4496 - val_loss: 1.5992 - val_accuracy: 0.4943 - 238ms/epoch - 5ms/step
Epoch 18/300
46/46 - 0s - loss: 1.7701 - accuracy: 0.4505 - val_loss: 1.5676 - val_accuracy: 0.5047 - 241ms/epoch - 5ms/step
Epoch 19/300
46/46 - 0s - loss: 1.7424 - accuracy: 0.4526 - val_loss: 1.5745 - val_accuracy: 0.4803 - 240ms/epoch - 5ms/step
Epoch 20/300
46/46 - 0s - loss: 1.7054 - accuracy: 0.4730 - val_loss: 1.4673 - val_accuracy: 0.5347 - 240ms/epoch - 5ms/step
Epoch 21/300
46/46 - 0s - loss: 1.6427 - accuracy: 0.4832 - val_loss: 1.4325 - val_accuracy: 0.5460 - 236ms/epoch - 5ms/step
Epoch 22/300
46/46 - 0s - loss: 1.6460 - accuracy: 0.4833 - val_loss: 1.4250 - val_accuracy: 0.5413 - 241ms/epoch - 5ms/step
Epoch 23/300
46/46 - 0s - loss: 1.6023 - accuracy: 0.4945 - val_loss: 1.4021 - val_accuracy: 0.5503 - 242ms/epoch - 5ms/step
Epoch 24/300
46/46 - 0s - loss: 1.6037 - accuracy: 0.4987 - val_loss: 1.4064 - val_accuracy: 0.5560 - 239ms/epoch - 5ms/step
Epoch 25/300
46/46 - 0s - loss: 1.5658 - accuracy: 0.5055 - val_loss: 1.4155 - val_accuracy: 0.5410 - 239ms/epoch - 5ms/step
Epoch 26/300
46/46 - 0s - loss: 1.5735 - accuracy: 0.5100 - val_loss: 1.3408 - val_accuracy: 0.5797 - 243ms/epoch - 5ms/step
Epoch 27/300
46/46 - 0s - loss: 1.5390 - accuracy: 0.5183 - val_loss: 1.3584 - val_accuracy: 0.5740 - 242ms/epoch - 5ms/step
Epoch 28/300
46/46 - 0s - loss: 1.5081 - accuracy: 0.5276 - val_loss: 1.3591 - val_accuracy: 0.5590 - 240ms/epoch - 5ms/step
Epoch 29/300
46/46 - 0s - loss: 1.5069 - accuracy: 0.5258 - val_loss: 1.2855 - val_accuracy: 0.6097 - 238ms/epoch - 5ms/step
Epoch 30/300
46/46 - 0s - loss: 1.4766 - accuracy: 0.5407 - val_loss: 1.2636 - val_accuracy: 0.6087 - 242ms/epoch - 5ms/step
Epoch 31/300
46/46 - 0s - loss: 1.4698 - accuracy: 0.5394 - val_loss: 1.3152 - val_accuracy: 0.5943 - 252ms/epoch - 5ms/step
Epoch 32/300
46/46 - 0s - loss: 1.4684 - accuracy: 0.5381 - val_loss: 1.3086 - val_accuracy: 0.5847 - 254ms/epoch - 6ms/step
Epoch 33/300
46/46 - 0s - loss: 1.4574 - accuracy: 0.5450 - val_loss: 1.2305 - val_accuracy: 0.6050 - 259ms/epoch - 6ms/step
Epoch 34/300
46/46 - 0s - loss: 1.4373 - accuracy: 0.5505 - val_loss: 1.1955 - val_accuracy: 0.6377 - 291ms/epoch - 6ms/step
Epoch 35/300
46/46 - 0s - loss: 1.4157 - accuracy: 0.5573 - val_loss: 1.2175 - val_accuracy: 0.6117 - 271ms/epoch - 6ms/step
Epoch 36/300
46/46 - 0s - loss: 1.3824 - accuracy: 0.5691 - val_loss: 1.2042 - val_accuracy: 0.6283 - 272ms/epoch - 6ms/step
Epoch 37/300
46/46 - 0s - loss: 1.3953 - accuracy: 0.5713 - val_loss: 1.1963 - val_accuracy: 0.6207 - 252ms/epoch - 5ms/step
Epoch 38/300
46/46 - 0s - loss: 1.3868 - accuracy: 0.5671 - val_loss: 1.2475 - val_accuracy: 0.6300 - 259ms/epoch - 6ms/step
Epoch 39/300
46/46 - 0s - loss: 1.3602 - accuracy: 0.5749 - val_loss: 1.1489 - val_accuracy: 0.6343 - 253ms/epoch - 6ms/step
Epoch 40/300
46/46 - 0s - loss: 1.3477 - accuracy: 0.5755 - val_loss: 1.1954 - val_accuracy: 0.6227 - 245ms/epoch - 5ms/step
Epoch 41/300
46/46 - 0s - loss: 1.3312 - accuracy: 0.5831 - val_loss: 1.1082 - val_accuracy: 0.6460 - 255ms/epoch - 6ms/step
Epoch 42/300
46/46 - 0s - loss: 1.3150 - accuracy: 0.5843 - val_loss: 1.1305 - val_accuracy: 0.6467 - 249ms/epoch - 5ms/step
Epoch 43/300
46/46 - 0s - loss: 1.3277 - accuracy: 0.5879 - val_loss: 1.1050 - val_accuracy: 0.6697 - 240ms/epoch - 5ms/step
Epoch 44/300
46/46 - 0s - loss: 1.2894 - accuracy: 0.5950 - val_loss: 1.0945 - val_accuracy: 0.6600 - 236ms/epoch - 5ms/step
Epoch 45/300
46/46 - 0s - loss: 1.3236 - accuracy: 0.5879 - val_loss: 1.1060 - val_accuracy: 0.6560 - 238ms/epoch - 5ms/step
Epoch 46/300
46/46 - 0s - loss: 1.3017 - accuracy: 0.6043 - val_loss: 1.0384 - val_accuracy: 0.6857 - 248ms/epoch - 5ms/step
Epoch 47/300
46/46 - 0s - loss: 1.2585 - accuracy: 0.6095 - val_loss: 1.0386 - val_accuracy: 0.6800 - 241ms/epoch - 5ms/step
Epoch 48/300
46/46 - 0s - loss: 1.2733 - accuracy: 0.6062 - val_loss: 1.0364 - val_accuracy: 0.6847 - 238ms/epoch - 5ms/step
Epoch 49/300
46/46 - 0s - loss: 1.2632 - accuracy: 0.6038 - val_loss: 1.0420 - val_accuracy: 0.6870 - 241ms/epoch - 5ms/step
Epoch 50/300
46/46 - 0s - loss: 1.2482 - accuracy: 0.6135 - val_loss: 1.0327 - val_accuracy: 0.6927 - 242ms/epoch - 5ms/step
Epoch 51/300
46/46 - 0s - loss: 1.2764 - accuracy: 0.6117 - val_loss: 1.3639 - val_accuracy: 0.5673 - 251ms/epoch - 5ms/step
Epoch 52/300
46/46 - 0s - loss: 1.2663 - accuracy: 0.6070 - val_loss: 0.9944 - val_accuracy: 0.6920 - 242ms/epoch - 5ms/step
Epoch 53/300
46/46 - 0s - loss: 1.2330 - accuracy: 0.6194 - val_loss: 0.9808 - val_accuracy: 0.7053 - 245ms/epoch - 5ms/step
Epoch 54/300
46/46 - 0s - loss: 1.2001 - accuracy: 0.6275 - val_loss: 0.9920 - val_accuracy: 0.7037 - 245ms/epoch - 5ms/step
Epoch 55/300
46/46 - 0s - loss: 1.1951 - accuracy: 0.6251 - val_loss: 1.0116 - val_accuracy: 0.6797 - 243ms/epoch - 5ms/step
Epoch 56/300
46/46 - 0s - loss: 1.1617 - accuracy: 0.6449 - val_loss: 0.9666 - val_accuracy: 0.7110 - 245ms/epoch - 5ms/step
Epoch 57/300
46/46 - 0s - loss: 1.1791 - accuracy: 0.6346 - val_loss: 1.0042 - val_accuracy: 0.6910 - 244ms/epoch - 5ms/step
Epoch 58/300
46/46 - 0s - loss: 1.1887 - accuracy: 0.6313 - val_loss: 0.9660 - val_accuracy: 0.7160 - 245ms/epoch - 5ms/step
Epoch 59/300
46/46 - 0s - loss: 1.1641 - accuracy: 0.6380 - val_loss: 0.9585 - val_accuracy: 0.7037 - 242ms/epoch - 5ms/step
Epoch 60/300
46/46 - 0s - loss: 1.1951 - accuracy: 0.6314 - val_loss: 1.0038 - val_accuracy: 0.7090 - 240ms/epoch - 5ms/step
Epoch 61/300
46/46 - 0s - loss: 1.1488 - accuracy: 0.6448 - val_loss: 0.9085 - val_accuracy: 0.7260 - 237ms/epoch - 5ms/step
Epoch 62/300
46/46 - 0s - loss: 1.1471 - accuracy: 0.6499 - val_loss: 0.9097 - val_accuracy: 0.7293 - 244ms/epoch - 5ms/step
Epoch 63/300
46/46 - 0s - loss: 1.1719 - accuracy: 0.6413 - val_loss: 0.9388 - val_accuracy: 0.7227 - 247ms/epoch - 5ms/step
Epoch 64/300
46/46 - 0s - loss: 1.1391 - accuracy: 0.6488 - val_loss: 0.9802 - val_accuracy: 0.6997 - 240ms/epoch - 5ms/step
Epoch 65/300
46/46 - 0s - loss: 1.1279 - accuracy: 0.6493 - val_loss: 0.8956 - val_accuracy: 0.7283 - 244ms/epoch - 5ms/step
Epoch 66/300
46/46 - 0s - loss: 1.1319 - accuracy: 0.6498 - val_loss: 0.9588 - val_accuracy: 0.7063 - 249ms/epoch - 5ms/step
Epoch 67/300
46/46 - 0s - loss: 1.1065 - accuracy: 0.6576 - val_loss: 0.9161 - val_accuracy: 0.7210 - 245ms/epoch - 5ms/step
Epoch 68/300
46/46 - 0s - loss: 1.1178 - accuracy: 0.6555 - val_loss: 0.8721 - val_accuracy: 0.7387 - 240ms/epoch - 5ms/step
Epoch 69/300
46/46 - 0s - loss: 1.0927 - accuracy: 0.6611 - val_loss: 0.8943 - val_accuracy: 0.7343 - 240ms/epoch - 5ms/step
Epoch 70/300
46/46 - 0s - loss: 1.0763 - accuracy: 0.6675 - val_loss: 0.9083 - val_accuracy: 0.7203 - 248ms/epoch - 5ms/step
Epoch 71/300
46/46 - 0s - loss: 1.0705 - accuracy: 0.6578 - val_loss: 0.8815 - val_accuracy: 0.7443 - 244ms/epoch - 5ms/step
Epoch 72/300
46/46 - 0s - loss: 1.0701 - accuracy: 0.6666 - val_loss: 0.8490 - val_accuracy: 0.7503 - 243ms/epoch - 5ms/step
Epoch 73/300
46/46 - 0s - loss: 1.0871 - accuracy: 0.6712 - val_loss: 0.8685 - val_accuracy: 0.7393 - 242ms/epoch - 5ms/step
Epoch 74/300
46/46 - 0s - loss: 1.0632 - accuracy: 0.6710 - val_loss: 0.8806 - val_accuracy: 0.7360 - 244ms/epoch - 5ms/step
Epoch 75/300
46/46 - 0s - loss: 1.0507 - accuracy: 0.6729 - val_loss: 0.9249 - val_accuracy: 0.7200 - 238ms/epoch - 5ms/step
Epoch 76/300
46/46 - 0s - loss: 1.1020 - accuracy: 0.6584 - val_loss: 0.8573 - val_accuracy: 0.7420 - 241ms/epoch - 5ms/step
Epoch 77/300
46/46 - 0s - loss: 1.0304 - accuracy: 0.6793 - val_loss: 0.8452 - val_accuracy: 0.7480 - 238ms/epoch - 5ms/step
Epoch 78/300
46/46 - 0s - loss: 1.0487 - accuracy: 0.6799 - val_loss: 0.8666 - val_accuracy: 0.7433 - 246ms/epoch - 5ms/step
Epoch 79/300
46/46 - 0s - loss: 1.0663 - accuracy: 0.6674 - val_loss: 0.8462 - val_accuracy: 0.7493 - 241ms/epoch - 5ms/step
Epoch 80/300
46/46 - 0s - loss: 1.0278 - accuracy: 0.6899 - val_loss: 0.8399 - val_accuracy: 0.7523 - 237ms/epoch - 5ms/step
Epoch 81/300
46/46 - 0s - loss: 1.0188 - accuracy: 0.6766 - val_loss: 0.8467 - val_accuracy: 0.7363 - 241ms/epoch - 5ms/step
Epoch 82/300
46/46 - 0s - loss: 1.0241 - accuracy: 0.6809 - val_loss: 0.9900 - val_accuracy: 0.6900 - 242ms/epoch - 5ms/step
Epoch 83/300
46/46 - 0s - loss: 1.0541 - accuracy: 0.6799 - val_loss: 0.8056 - val_accuracy: 0.7530 - 245ms/epoch - 5ms/step
Epoch 84/300
46/46 - 0s - loss: 1.0107 - accuracy: 0.6883 - val_loss: 0.8049 - val_accuracy: 0.7567 - 242ms/epoch - 5ms/step
Epoch 85/300
46/46 - 0s - loss: 1.0126 - accuracy: 0.6832 - val_loss: 0.7989 - val_accuracy: 0.7567 - 239ms/epoch - 5ms/step
Epoch 86/300
46/46 - 0s - loss: 0.9940 - accuracy: 0.6918 - val_loss: 0.7892 - val_accuracy: 0.7657 - 242ms/epoch - 5ms/step
Epoch 87/300
46/46 - 0s - loss: 0.9943 - accuracy: 0.6954 - val_loss: 0.7946 - val_accuracy: 0.7663 - 244ms/epoch - 5ms/step
Epoch 88/300
46/46 - 0s - loss: 1.0075 - accuracy: 0.6827 - val_loss: 0.7905 - val_accuracy: 0.7647 - 239ms/epoch - 5ms/step
Epoch 89/300
46/46 - 0s - loss: 1.0159 - accuracy: 0.6936 - val_loss: 0.8179 - val_accuracy: 0.7490 - 237ms/epoch - 5ms/step
Epoch 90/300
46/46 - 0s - loss: 1.0025 - accuracy: 0.6901 - val_loss: 0.8132 - val_accuracy: 0.7630 - 241ms/epoch - 5ms/step
Epoch 91/300
46/46 - 0s - loss: 0.9852 - accuracy: 0.6952 - val_loss: 0.7740 - val_accuracy: 0.7740 - 242ms/epoch - 5ms/step
Epoch 92/300
46/46 - 0s - loss: 0.9770 - accuracy: 0.6972 - val_loss: 0.7843 - val_accuracy: 0.7630 - 243ms/epoch - 5ms/step
Epoch 93/300
46/46 - 0s - loss: 0.9607 - accuracy: 0.7023 - val_loss: 0.8402 - val_accuracy: 0.7463 - 245ms/epoch - 5ms/step
Epoch 94/300
46/46 - 0s - loss: 0.9963 - accuracy: 0.6948 - val_loss: 0.7753 - val_accuracy: 0.7747 - 239ms/epoch - 5ms/step
Epoch 95/300
46/46 - 0s - loss: 0.9638 - accuracy: 0.7015 - val_loss: 0.8501 - val_accuracy: 0.7393 - 242ms/epoch - 5ms/step
Epoch 96/300
46/46 - 0s - loss: 0.9647 - accuracy: 0.7043 - val_loss: 0.7738 - val_accuracy: 0.7753 - 241ms/epoch - 5ms/step
Epoch 97/300
46/46 - 0s - loss: 0.9663 - accuracy: 0.7010 - val_loss: 0.7420 - val_accuracy: 0.7860 - 245ms/epoch - 5ms/step
Epoch 98/300
46/46 - 0s - loss: 0.9459 - accuracy: 0.7117 - val_loss: 0.7502 - val_accuracy: 0.7747 - 243ms/epoch - 5ms/step
Epoch 99/300
46/46 - 0s - loss: 0.9256 - accuracy: 0.7136 - val_loss: 0.7598 - val_accuracy: 0.7690 - 243ms/epoch - 5ms/step
Epoch 100/300
46/46 - 0s - loss: 0.9447 - accuracy: 0.7130 - val_loss: 0.9506 - val_accuracy: 0.7053 - 248ms/epoch - 5ms/step
Epoch 101/300
46/46 - 0s - loss: 0.9506 - accuracy: 0.7040 - val_loss: 0.7909 - val_accuracy: 0.7677 - 239ms/epoch - 5ms/step
Epoch 102/300
46/46 - 0s - loss: 0.9321 - accuracy: 0.7126 - val_loss: 0.7986 - val_accuracy: 0.7643 - 243ms/epoch - 5ms/step
Epoch 103/300
46/46 - 0s - loss: 0.9509 - accuracy: 0.7091 - val_loss: 0.7411 - val_accuracy: 0.7803 - 244ms/epoch - 5ms/step
Epoch 104/300
46/46 - 0s - loss: 0.9028 - accuracy: 0.7179 - val_loss: 0.7225 - val_accuracy: 0.7883 - 249ms/epoch - 5ms/step
Epoch 105/300
46/46 - 0s - loss: 0.9187 - accuracy: 0.7141 - val_loss: 0.7278 - val_accuracy: 0.7857 - 242ms/epoch - 5ms/step
Epoch 106/300
46/46 - 0s - loss: 0.9108 - accuracy: 0.7137 - val_loss: 0.7084 - val_accuracy: 0.7923 - 243ms/epoch - 5ms/step
Epoch 107/300
46/46 - 0s - loss: 0.9165 - accuracy: 0.7172 - val_loss: 0.7123 - val_accuracy: 0.7877 - 243ms/epoch - 5ms/step
Epoch 108/300
46/46 - 0s - loss: 0.9156 - accuracy: 0.7151 - val_loss: 0.7155 - val_accuracy: 0.7883 - 241ms/epoch - 5ms/step
Epoch 109/300
46/46 - 0s - loss: 0.9082 - accuracy: 0.7178 - val_loss: 0.7296 - val_accuracy: 0.7827 - 239ms/epoch - 5ms/step
Epoch 110/300
46/46 - 0s - loss: 0.9052 - accuracy: 0.7213 - val_loss: 0.7066 - val_accuracy: 0.7893 - 240ms/epoch - 5ms/step
Epoch 111/300
46/46 - 0s - loss: 0.9065 - accuracy: 0.7151 - val_loss: 0.7430 - val_accuracy: 0.7760 - 251ms/epoch - 5ms/step
Epoch 112/300
46/46 - 0s - loss: 0.8839 - accuracy: 0.7292 - val_loss: 0.6939 - val_accuracy: 0.7937 - 253ms/epoch - 6ms/step
Epoch 113/300
46/46 - 0s - loss: 0.8820 - accuracy: 0.7277 - val_loss: 0.7625 - val_accuracy: 0.7727 - 249ms/epoch - 5ms/step
Epoch 114/300
46/46 - 0s - loss: 0.8845 - accuracy: 0.7280 - val_loss: 0.7212 - val_accuracy: 0.7803 - 247ms/epoch - 5ms/step
Epoch 115/300
46/46 - 0s - loss: 0.8947 - accuracy: 0.7297 - val_loss: 0.7205 - val_accuracy: 0.7893 - 242ms/epoch - 5ms/step
Epoch 116/300
46/46 - 0s - loss: 0.8618 - accuracy: 0.7353 - val_loss: 0.7030 - val_accuracy: 0.7930 - 245ms/epoch - 5ms/step
Epoch 117/300
46/46 - 0s - loss: 0.8528 - accuracy: 0.7352 - val_loss: 0.6832 - val_accuracy: 0.7967 - 254ms/epoch - 6ms/step
Epoch 118/300
46/46 - 0s - loss: 0.8455 - accuracy: 0.7410 - val_loss: 0.7073 - val_accuracy: 0.7913 - 255ms/epoch - 6ms/step
Epoch 119/300
46/46 - 0s - loss: 0.8373 - accuracy: 0.7441 - val_loss: 0.7050 - val_accuracy: 0.7837 - 255ms/epoch - 6ms/step
Epoch 120/300
46/46 - 0s - loss: 0.8384 - accuracy: 0.7441 - val_loss: 0.6915 - val_accuracy: 0.7933 - 248ms/epoch - 5ms/step
Epoch 121/300
46/46 - 0s - loss: 0.8435 - accuracy: 0.7314 - val_loss: 0.6492 - val_accuracy: 0.8093 - 238ms/epoch - 5ms/step
Epoch 122/300
46/46 - 0s - loss: 0.8544 - accuracy: 0.7365 - val_loss: 0.6849 - val_accuracy: 0.8000 - 241ms/epoch - 5ms/step
Epoch 123/300
46/46 - 0s - loss: 0.9212 - accuracy: 0.7171 - val_loss: 0.6738 - val_accuracy: 0.7990 - 251ms/epoch - 5ms/step
Epoch 124/300
46/46 - 0s - loss: 0.8714 - accuracy: 0.7326 - val_loss: 0.6655 - val_accuracy: 0.8070 - 253ms/epoch - 5ms/step
Epoch 125/300
46/46 - 0s - loss: 0.8296 - accuracy: 0.7368 - val_loss: 0.6863 - val_accuracy: 0.7987 - 248ms/epoch - 5ms/step
Epoch 126/300
46/46 - 0s - loss: 0.8510 - accuracy: 0.7340 - val_loss: 0.6718 - val_accuracy: 0.8050 - 246ms/epoch - 5ms/step
Epoch 127/300
46/46 - 0s - loss: 0.8244 - accuracy: 0.7477 - val_loss: 0.7033 - val_accuracy: 0.7903 - 249ms/epoch - 5ms/step
Epoch 128/300
46/46 - 0s - loss: 0.8210 - accuracy: 0.7471 - val_loss: 0.6675 - val_accuracy: 0.7993 - 242ms/epoch - 5ms/step
Epoch 129/300
46/46 - 0s - loss: 0.8109 - accuracy: 0.7490 - val_loss: 0.6925 - val_accuracy: 0.7890 - 240ms/epoch - 5ms/step
Epoch 130/300
46/46 - 0s - loss: 0.8227 - accuracy: 0.7462 - val_loss: 0.6989 - val_accuracy: 0.7913 - 241ms/epoch - 5ms/step
Epoch 131/300
46/46 - 0s - loss: 0.8080 - accuracy: 0.7478 - val_loss: 0.6692 - val_accuracy: 0.8020 - 242ms/epoch - 5ms/step
Epoch 132/300
46/46 - 0s - loss: 0.8006 - accuracy: 0.7542 - val_loss: 0.6565 - val_accuracy: 0.8043 - 242ms/epoch - 5ms/step
Epoch 133/300
46/46 - 0s - loss: 0.8290 - accuracy: 0.7414 - val_loss: 0.6308 - val_accuracy: 0.8140 - 244ms/epoch - 5ms/step
Epoch 134/300
46/46 - 0s - loss: 0.8036 - accuracy: 0.7511 - val_loss: 0.6451 - val_accuracy: 0.8093 - 243ms/epoch - 5ms/step
Epoch 135/300
46/46 - 0s - loss: 0.7906 - accuracy: 0.7555 - val_loss: 0.6182 - val_accuracy: 0.8157 - 243ms/epoch - 5ms/step
Epoch 136/300
46/46 - 0s - loss: 0.7862 - accuracy: 0.7578 - val_loss: 0.6409 - val_accuracy: 0.8050 - 243ms/epoch - 5ms/step
Epoch 137/300
46/46 - 0s - loss: 0.7840 - accuracy: 0.7576 - val_loss: 0.6814 - val_accuracy: 0.7970 - 241ms/epoch - 5ms/step
Epoch 138/300
46/46 - 0s - loss: 0.8038 - accuracy: 0.7535 - val_loss: 0.6252 - val_accuracy: 0.8180 - 241ms/epoch - 5ms/step
Epoch 139/300
46/46 - 0s - loss: 0.7951 - accuracy: 0.7523 - val_loss: 0.6206 - val_accuracy: 0.8167 - 246ms/epoch - 5ms/step
Epoch 140/300
46/46 - 0s - loss: 0.7933 - accuracy: 0.7549 - val_loss: 0.6261 - val_accuracy: 0.8123 - 240ms/epoch - 5ms/step
Epoch 141/300
46/46 - 0s - loss: 0.7570 - accuracy: 0.7647 - val_loss: 0.6358 - val_accuracy: 0.8120 - 239ms/epoch - 5ms/step
Epoch 142/300
46/46 - 0s - loss: 0.7899 - accuracy: 0.7532 - val_loss: 0.6334 - val_accuracy: 0.8163 - 238ms/epoch - 5ms/step
Epoch 143/300
46/46 - 0s - loss: 0.7588 - accuracy: 0.7650 - val_loss: 0.6075 - val_accuracy: 0.8193 - 244ms/epoch - 5ms/step
Epoch 144/300
46/46 - 0s - loss: 0.7627 - accuracy: 0.7632 - val_loss: 0.6063 - val_accuracy: 0.8187 - 238ms/epoch - 5ms/step
Epoch 145/300
46/46 - 0s - loss: 0.7646 - accuracy: 0.7692 - val_loss: 0.6031 - val_accuracy: 0.8273 - 241ms/epoch - 5ms/step
Epoch 146/300
46/46 - 0s - loss: 0.7741 - accuracy: 0.7633 - val_loss: 0.6102 - val_accuracy: 0.8213 - 240ms/epoch - 5ms/step
Epoch 147/300
46/46 - 0s - loss: 0.7532 - accuracy: 0.7620 - val_loss: 0.5937 - val_accuracy: 0.8273 - 246ms/epoch - 5ms/step
Epoch 148/300
46/46 - 0s - loss: 0.7657 - accuracy: 0.7643 - val_loss: 0.6118 - val_accuracy: 0.8163 - 244ms/epoch - 5ms/step
Epoch 149/300
46/46 - 0s - loss: 0.7584 - accuracy: 0.7657 - val_loss: 0.6185 - val_accuracy: 0.8190 - 247ms/epoch - 5ms/step
Epoch 150/300
46/46 - 0s - loss: 0.7406 - accuracy: 0.7736 - val_loss: 0.5850 - val_accuracy: 0.8313 - 241ms/epoch - 5ms/step
Epoch 151/300
46/46 - 0s - loss: 0.7458 - accuracy: 0.7700 - val_loss: 0.6473 - val_accuracy: 0.8100 - 245ms/epoch - 5ms/step
Epoch 152/300
46/46 - 0s - loss: 0.7441 - accuracy: 0.7692 - val_loss: 0.5679 - val_accuracy: 0.8327 - 252ms/epoch - 5ms/step
Epoch 153/300
46/46 - 0s - loss: 0.7335 - accuracy: 0.7768 - val_loss: 0.5870 - val_accuracy: 0.8290 - 245ms/epoch - 5ms/step
Epoch 154/300
46/46 - 0s - loss: 0.7127 - accuracy: 0.7799 - val_loss: 0.6236 - val_accuracy: 0.8190 - 240ms/epoch - 5ms/step
Epoch 155/300
46/46 - 0s - loss: 0.7495 - accuracy: 0.7669 - val_loss: 0.5826 - val_accuracy: 0.8307 - 244ms/epoch - 5ms/step
Epoch 156/300
46/46 - 0s - loss: 0.7221 - accuracy: 0.7807 - val_loss: 0.5977 - val_accuracy: 0.8267 - 245ms/epoch - 5ms/step
Epoch 157/300
46/46 - 0s - loss: 0.7071 - accuracy: 0.7807 - val_loss: 0.7105 - val_accuracy: 0.7837 - 249ms/epoch - 5ms/step
Epoch 158/300
46/46 - 0s - loss: 0.7548 - accuracy: 0.7645 - val_loss: 0.6237 - val_accuracy: 0.8197 - 249ms/epoch - 5ms/step
Epoch 159/300
46/46 - 0s - loss: 0.7183 - accuracy: 0.7734 - val_loss: 0.5746 - val_accuracy: 0.8350 - 257ms/epoch - 6ms/step
Epoch 160/300
46/46 - 0s - loss: 0.7232 - accuracy: 0.7769 - val_loss: 0.5718 - val_accuracy: 0.8350 - 247ms/epoch - 5ms/step
Epoch 161/300
46/46 - 0s - loss: 0.7119 - accuracy: 0.7799 - val_loss: 0.5788 - val_accuracy: 0.8357 - 243ms/epoch - 5ms/step
Epoch 162/300
46/46 - 0s - loss: 0.7412 - accuracy: 0.7728 - val_loss: 0.5712 - val_accuracy: 0.8353 - 243ms/epoch - 5ms/step
Epoch 163/300
46/46 - 0s - loss: 0.7186 - accuracy: 0.7750 - val_loss: 0.5637 - val_accuracy: 0.8357 - 248ms/epoch - 5ms/step
Epoch 164/300
46/46 - 0s - loss: 0.7055 - accuracy: 0.7897 - val_loss: 0.6074 - val_accuracy: 0.8243 - 244ms/epoch - 5ms/step
Epoch 165/300
46/46 - 0s - loss: 0.6865 - accuracy: 0.7853 - val_loss: 0.5989 - val_accuracy: 0.8197 - 251ms/epoch - 5ms/step
Epoch 166/300
46/46 - 0s - loss: 0.7205 - accuracy: 0.7767 - val_loss: 0.5606 - val_accuracy: 0.8367 - 246ms/epoch - 5ms/step
Epoch 167/300
46/46 - 0s - loss: 0.7012 - accuracy: 0.7809 - val_loss: 0.5527 - val_accuracy: 0.8400 - 251ms/epoch - 5ms/step
Epoch 168/300
46/46 - 0s - loss: 0.6978 - accuracy: 0.7851 - val_loss: 0.5528 - val_accuracy: 0.8383 - 248ms/epoch - 5ms/step
Epoch 169/300
46/46 - 0s - loss: 0.6989 - accuracy: 0.7870 - val_loss: 0.6211 - val_accuracy: 0.8177 - 245ms/epoch - 5ms/step
Epoch 170/300
46/46 - 0s - loss: 0.7037 - accuracy: 0.7833 - val_loss: 0.5554 - val_accuracy: 0.8457 - 242ms/epoch - 5ms/step
Epoch 171/300
46/46 - 0s - loss: 0.6935 - accuracy: 0.7894 - val_loss: 0.5818 - val_accuracy: 0.8257 - 246ms/epoch - 5ms/step
Epoch 172/300
46/46 - 0s - loss: 0.6856 - accuracy: 0.7911 - val_loss: 0.5757 - val_accuracy: 0.8297 - 243ms/epoch - 5ms/step
Epoch 173/300
46/46 - 0s - loss: 0.7003 - accuracy: 0.7907 - val_loss: 0.5923 - val_accuracy: 0.8270 - 241ms/epoch - 5ms/step
Epoch 174/300
46/46 - 0s - loss: 0.6728 - accuracy: 0.7910 - val_loss: 0.6113 - val_accuracy: 0.8283 - 242ms/epoch - 5ms/step
Epoch 175/300
46/46 - 0s - loss: 0.7024 - accuracy: 0.7862 - val_loss: 0.6084 - val_accuracy: 0.8167 - 246ms/epoch - 5ms/step
Epoch 176/300
46/46 - 0s - loss: 0.6896 - accuracy: 0.7928 - val_loss: 0.5532 - val_accuracy: 0.8450 - 245ms/epoch - 5ms/step
Epoch 177/300
46/46 - 0s - loss: 0.6659 - accuracy: 0.7912 - val_loss: 0.5269 - val_accuracy: 0.8487 - 243ms/epoch - 5ms/step
Epoch 178/300
46/46 - 0s - loss: 0.6914 - accuracy: 0.7903 - val_loss: 0.6001 - val_accuracy: 0.8243 - 249ms/epoch - 5ms/step
Epoch 179/300
46/46 - 0s - loss: 0.6691 - accuracy: 0.7951 - val_loss: 0.5893 - val_accuracy: 0.8267 - 250ms/epoch - 5ms/step
Epoch 180/300
46/46 - 0s - loss: 0.6583 - accuracy: 0.7980 - val_loss: 0.5442 - val_accuracy: 0.8433 - 242ms/epoch - 5ms/step
Epoch 181/300
46/46 - 0s - loss: 0.6621 - accuracy: 0.7961 - val_loss: 0.5334 - val_accuracy: 0.8487 - 248ms/epoch - 5ms/step
Epoch 182/300
46/46 - 0s - loss: 0.6534 - accuracy: 0.7979 - val_loss: 0.5325 - val_accuracy: 0.8457 - 240ms/epoch - 5ms/step
Epoch 183/300
46/46 - 0s - loss: 0.6492 - accuracy: 0.8011 - val_loss: 0.5408 - val_accuracy: 0.8450 - 245ms/epoch - 5ms/step
Epoch 184/300
46/46 - 0s - loss: 0.6388 - accuracy: 0.8034 - val_loss: 0.5244 - val_accuracy: 0.8510 - 245ms/epoch - 5ms/step
Epoch 185/300
46/46 - 0s - loss: 0.6470 - accuracy: 0.8057 - val_loss: 0.5179 - val_accuracy: 0.8517 - 252ms/epoch - 5ms/step
Epoch 186/300
46/46 - 0s - loss: 0.6620 - accuracy: 0.7935 - val_loss: 0.5370 - val_accuracy: 0.8483 - 247ms/epoch - 5ms/step
Epoch 187/300
46/46 - 0s - loss: 0.6397 - accuracy: 0.8048 - val_loss: 0.5684 - val_accuracy: 0.8363 - 249ms/epoch - 5ms/step
Epoch 188/300
46/46 - 0s - loss: 0.6288 - accuracy: 0.8044 - val_loss: 0.5292 - val_accuracy: 0.8510 - 256ms/epoch - 6ms/step
Epoch 189/300
46/46 - 0s - loss: 0.6461 - accuracy: 0.8053 - val_loss: 0.5845 - val_accuracy: 0.8300 - 244ms/epoch - 5ms/step
Epoch 190/300
46/46 - 0s - loss: 0.6512 - accuracy: 0.8004 - val_loss: 0.5288 - val_accuracy: 0.8477 - 241ms/epoch - 5ms/step
Epoch 191/300
46/46 - 0s - loss: 0.6279 - accuracy: 0.8062 - val_loss: 0.5256 - val_accuracy: 0.8517 - 247ms/epoch - 5ms/step
Epoch 192/300
46/46 - 0s - loss: 0.6468 - accuracy: 0.8015 - val_loss: 0.5444 - val_accuracy: 0.8437 - 246ms/epoch - 5ms/step
Epoch 193/300
46/46 - 0s - loss: 0.6523 - accuracy: 0.8047 - val_loss: 0.5569 - val_accuracy: 0.8387 - 245ms/epoch - 5ms/step
Epoch 194/300
46/46 - 0s - loss: 0.6447 - accuracy: 0.7969 - val_loss: 0.5404 - val_accuracy: 0.8523 - 243ms/epoch - 5ms/step
Epoch 195/300
46/46 - 0s - loss: 0.6089 - accuracy: 0.8108 - val_loss: 0.5009 - val_accuracy: 0.8567 - 244ms/epoch - 5ms/step
Epoch 196/300
46/46 - 0s - loss: 0.6279 - accuracy: 0.8117 - val_loss: 0.4986 - val_accuracy: 0.8597 - 243ms/epoch - 5ms/step
Epoch 197/300
46/46 - 0s - loss: 0.6296 - accuracy: 0.8033 - val_loss: 0.5549 - val_accuracy: 0.8470 - 241ms/epoch - 5ms/step
Epoch 198/300
46/46 - 0s - loss: 0.6307 - accuracy: 0.8037 - val_loss: 0.5006 - val_accuracy: 0.8587 - 242ms/epoch - 5ms/step
Epoch 199/300
46/46 - 0s - loss: 0.6318 - accuracy: 0.8073 - val_loss: 0.4881 - val_accuracy: 0.8650 - 247ms/epoch - 5ms/step
Epoch 200/300
46/46 - 0s - loss: 0.6113 - accuracy: 0.8108 - val_loss: 0.5293 - val_accuracy: 0.8490 - 244ms/epoch - 5ms/step
Epoch 201/300
46/46 - 0s - loss: 0.6245 - accuracy: 0.8069 - val_loss: 0.5279 - val_accuracy: 0.8437 - 243ms/epoch - 5ms/step
Epoch 202/300
46/46 - 0s - loss: 0.6068 - accuracy: 0.8141 - val_loss: 0.5217 - val_accuracy: 0.8503 - 241ms/epoch - 5ms/step
Epoch 203/300
46/46 - 0s - loss: 0.6096 - accuracy: 0.8114 - val_loss: 0.4902 - val_accuracy: 0.8630 - 243ms/epoch - 5ms/step
Epoch 204/300
46/46 - 0s - loss: 0.5950 - accuracy: 0.8156 - val_loss: 0.5283 - val_accuracy: 0.8513 - 247ms/epoch - 5ms/step
Epoch 205/300
46/46 - 0s - loss: 0.6305 - accuracy: 0.8082 - val_loss: 0.5056 - val_accuracy: 0.8540 - 241ms/epoch - 5ms/step
Epoch 206/300
46/46 - 0s - loss: 0.6122 - accuracy: 0.8121 - val_loss: 0.5068 - val_accuracy: 0.8597 - 242ms/epoch - 5ms/step
Epoch 207/300
46/46 - 0s - loss: 0.5979 - accuracy: 0.8154 - val_loss: 0.4966 - val_accuracy: 0.8597 - 247ms/epoch - 5ms/step
Epoch 208/300
46/46 - 0s - loss: 0.5900 - accuracy: 0.8165 - val_loss: 0.5139 - val_accuracy: 0.8547 - 243ms/epoch - 5ms/step
Epoch 209/300
46/46 - 0s - loss: 0.5925 - accuracy: 0.8201 - val_loss: 0.4859 - val_accuracy: 0.8650 - 245ms/epoch - 5ms/step
Epoch 210/300
46/46 - 0s - loss: 0.6076 - accuracy: 0.8162 - val_loss: 0.5349 - val_accuracy: 0.8517 - 243ms/epoch - 5ms/step
Epoch 211/300
46/46 - 0s - loss: 0.5971 - accuracy: 0.8204 - val_loss: 0.4874 - val_accuracy: 0.8633 - 243ms/epoch - 5ms/step
Epoch 212/300
46/46 - 0s - loss: 0.5955 - accuracy: 0.8238 - val_loss: 0.4922 - val_accuracy: 0.8643 - 241ms/epoch - 5ms/step
Epoch 213/300
46/46 - 0s - loss: 0.5941 - accuracy: 0.8195 - val_loss: 0.5090 - val_accuracy: 0.8567 - 242ms/epoch - 5ms/step
Epoch 214/300
46/46 - 0s - loss: 0.5981 - accuracy: 0.8161 - val_loss: 0.4757 - val_accuracy: 0.8670 - 242ms/epoch - 5ms/step
Epoch 215/300
46/46 - 0s - loss: 0.5775 - accuracy: 0.8230 - val_loss: 0.4870 - val_accuracy: 0.8650 - 242ms/epoch - 5ms/step
Epoch 216/300
46/46 - 0s - loss: 0.5821 - accuracy: 0.8160 - val_loss: 0.5028 - val_accuracy: 0.8543 - 242ms/epoch - 5ms/step
Epoch 217/300
46/46 - 0s - loss: 0.5847 - accuracy: 0.8188 - val_loss: 0.4892 - val_accuracy: 0.8670 - 239ms/epoch - 5ms/step
Epoch 218/300
46/46 - 0s - loss: 0.5788 - accuracy: 0.8247 - val_loss: 0.4770 - val_accuracy: 0.8673 - 244ms/epoch - 5ms/step
Epoch 219/300
46/46 - 0s - loss: 0.5898 - accuracy: 0.8219 - val_loss: 0.4866 - val_accuracy: 0.8640 - 244ms/epoch - 5ms/step
Epoch 220/300
46/46 - 0s - loss: 0.5926 - accuracy: 0.8180 - val_loss: 0.4894 - val_accuracy: 0.8623 - 243ms/epoch - 5ms/step
Epoch 221/300
46/46 - 0s - loss: 0.5850 - accuracy: 0.8165 - val_loss: 0.4731 - val_accuracy: 0.8700 - 249ms/epoch - 5ms/step
Epoch 222/300
46/46 - 0s - loss: 0.5829 - accuracy: 0.8213 - val_loss: 0.5524 - val_accuracy: 0.8420 - 262ms/epoch - 6ms/step
Epoch 223/300
46/46 - 0s - loss: 0.5977 - accuracy: 0.8196 - val_loss: 0.4955 - val_accuracy: 0.8587 - 263ms/epoch - 6ms/step
Epoch 224/300
46/46 - 0s - loss: 0.5672 - accuracy: 0.8219 - val_loss: 0.5151 - val_accuracy: 0.8537 - 256ms/epoch - 6ms/step
Epoch 225/300
46/46 - 0s - loss: 0.5960 - accuracy: 0.8170 - val_loss: 0.4802 - val_accuracy: 0.8677 - 256ms/epoch - 6ms/step
Epoch 226/300
46/46 - 0s - loss: 0.5690 - accuracy: 0.8301 - val_loss: 0.4740 - val_accuracy: 0.8673 - 253ms/epoch - 6ms/step
Epoch 227/300
46/46 - 0s - loss: 0.6034 - accuracy: 0.8190 - val_loss: 0.4942 - val_accuracy: 0.8697 - 253ms/epoch - 6ms/step
Epoch 228/300
46/46 - 0s - loss: 0.6007 - accuracy: 0.8166 - val_loss: 0.4844 - val_accuracy: 0.8630 - 244ms/epoch - 5ms/step
Epoch 229/300
46/46 - 0s - loss: 0.5360 - accuracy: 0.8335 - val_loss: 0.4755 - val_accuracy: 0.8630 - 246ms/epoch - 5ms/step
Epoch 230/300
46/46 - 0s - loss: 0.5804 - accuracy: 0.8211 - val_loss: 0.4778 - val_accuracy: 0.8720 - 241ms/epoch - 5ms/step
Epoch 231/300
46/46 - 0s - loss: 0.6041 - accuracy: 0.8175 - val_loss: 0.5459 - val_accuracy: 0.8447 - 239ms/epoch - 5ms/step
Epoch 232/300
46/46 - 0s - loss: 0.5662 - accuracy: 0.8203 - val_loss: 0.4830 - val_accuracy: 0.8660 - 249ms/epoch - 5ms/step
Epoch 233/300
46/46 - 0s - loss: 0.5675 - accuracy: 0.8291 - val_loss: 0.4618 - val_accuracy: 0.8747 - 248ms/epoch - 5ms/step
Epoch 234/300
46/46 - 0s - loss: 0.5575 - accuracy: 0.8306 - val_loss: 0.4718 - val_accuracy: 0.8703 - 249ms/epoch - 5ms/step
Epoch 235/300
46/46 - 0s - loss: 0.5671 - accuracy: 0.8299 - val_loss: 0.4727 - val_accuracy: 0.8733 - 243ms/epoch - 5ms/step
Epoch 236/300
46/46 - 0s - loss: 0.5725 - accuracy: 0.8319 - val_loss: 0.4844 - val_accuracy: 0.8637 - 244ms/epoch - 5ms/step
Epoch 237/300
46/46 - 0s - loss: 0.5786 - accuracy: 0.8296 - val_loss: 0.4669 - val_accuracy: 0.8717 - 243ms/epoch - 5ms/step
Epoch 238/300
46/46 - 0s - loss: 0.5437 - accuracy: 0.8283 - val_loss: 0.4912 - val_accuracy: 0.8677 - 240ms/epoch - 5ms/step
Epoch 239/300
46/46 - 0s - loss: 0.5475 - accuracy: 0.8323 - val_loss: 0.4647 - val_accuracy: 0.8757 - 244ms/epoch - 5ms/step
Epoch 240/300
46/46 - 0s - loss: 0.5684 - accuracy: 0.8255 - val_loss: 0.5044 - val_accuracy: 0.8677 - 242ms/epoch - 5ms/step
Epoch 241/300
46/46 - 0s - loss: 0.5371 - accuracy: 0.8335 - val_loss: 0.4586 - val_accuracy: 0.8757 - 248ms/epoch - 5ms/step
Epoch 242/300
46/46 - 0s - loss: 0.5663 - accuracy: 0.8290 - val_loss: 0.4442 - val_accuracy: 0.8770 - 253ms/epoch - 5ms/step
Epoch 243/300
46/46 - 0s - loss: 0.5283 - accuracy: 0.8412 - val_loss: 0.4687 - val_accuracy: 0.8657 - 262ms/epoch - 6ms/step
Epoch 244/300
46/46 - 0s - loss: 0.5486 - accuracy: 0.8345 - val_loss: 0.4621 - val_accuracy: 0.8733 - 259ms/epoch - 6ms/step
Epoch 245/300
46/46 - 0s - loss: 0.5488 - accuracy: 0.8301 - val_loss: 0.4509 - val_accuracy: 0.8753 - 243ms/epoch - 5ms/step
Epoch 246/300
46/46 - 0s - loss: 0.5423 - accuracy: 0.8352 - val_loss: 0.4677 - val_accuracy: 0.8710 - 239ms/epoch - 5ms/step
Epoch 247/300
46/46 - 0s - loss: 0.5394 - accuracy: 0.8326 - val_loss: 0.4474 - val_accuracy: 0.8800 - 244ms/epoch - 5ms/step
Epoch 248/300
46/46 - 0s - loss: 0.5434 - accuracy: 0.8354 - val_loss: 0.4726 - val_accuracy: 0.8680 - 245ms/epoch - 5ms/step
Epoch 249/300
46/46 - 0s - loss: 0.5605 - accuracy: 0.8307 - val_loss: 0.4920 - val_accuracy: 0.8650 - 246ms/epoch - 5ms/step
Epoch 250/300
46/46 - 0s - loss: 0.5165 - accuracy: 0.8388 - val_loss: 0.4492 - val_accuracy: 0.8797 - 241ms/epoch - 5ms/step
Epoch 251/300
46/46 - 0s - loss: 0.5182 - accuracy: 0.8432 - val_loss: 0.4375 - val_accuracy: 0.8763 - 240ms/epoch - 5ms/step
Epoch 252/300
46/46 - 0s - loss: 0.5287 - accuracy: 0.8387 - val_loss: 0.4394 - val_accuracy: 0.8820 - 244ms/epoch - 5ms/step
Epoch 253/300
46/46 - 0s - loss: 0.5129 - accuracy: 0.8385 - val_loss: 0.4918 - val_accuracy: 0.8633 - 243ms/epoch - 5ms/step
Epoch 254/300
46/46 - 0s - loss: 0.5350 - accuracy: 0.8396 - val_loss: 0.4434 - val_accuracy: 0.8790 - 241ms/epoch - 5ms/step
Epoch 255/300
46/46 - 0s - loss: 0.5332 - accuracy: 0.8392 - val_loss: 0.4747 - val_accuracy: 0.8723 - 240ms/epoch - 5ms/step
Epoch 256/300
46/46 - 0s - loss: 0.5443 - accuracy: 0.8340 - val_loss: 0.5203 - val_accuracy: 0.8553 - 243ms/epoch - 5ms/step
Epoch 257/300
46/46 - 0s - loss: 0.5563 - accuracy: 0.8335 - val_loss: 0.4900 - val_accuracy: 0.8663 - 241ms/epoch - 5ms/step
Epoch 258/300
46/46 - 0s - loss: 0.5161 - accuracy: 0.8455 - val_loss: 0.4608 - val_accuracy: 0.8763 - 249ms/epoch - 5ms/step
Epoch 259/300
46/46 - 0s - loss: 0.5617 - accuracy: 0.8272 - val_loss: 0.4827 - val_accuracy: 0.8727 - 247ms/epoch - 5ms/step
Epoch 260/300
46/46 - 0s - loss: 0.5340 - accuracy: 0.8366 - val_loss: 0.4630 - val_accuracy: 0.8683 - 247ms/epoch - 5ms/step
Epoch 261/300
46/46 - 0s - loss: 0.5105 - accuracy: 0.8487 - val_loss: 0.4552 - val_accuracy: 0.8803 - 243ms/epoch - 5ms/step
Epoch 262/300
46/46 - 0s - loss: 0.5039 - accuracy: 0.8433 - val_loss: 0.4705 - val_accuracy: 0.8667 - 240ms/epoch - 5ms/step
Epoch 263/300
46/46 - 0s - loss: 0.5401 - accuracy: 0.8354 - val_loss: 0.4493 - val_accuracy: 0.8790 - 242ms/epoch - 5ms/step
Epoch 264/300
46/46 - 0s - loss: 0.5153 - accuracy: 0.8397 - val_loss: 0.4637 - val_accuracy: 0.8773 - 239ms/epoch - 5ms/step
Epoch 265/300
46/46 - 0s - loss: 0.4926 - accuracy: 0.8439 - val_loss: 0.4271 - val_accuracy: 0.8853 - 240ms/epoch - 5ms/step
Epoch 266/300
46/46 - 0s - loss: 0.5210 - accuracy: 0.8459 - val_loss: 0.4450 - val_accuracy: 0.8820 - 241ms/epoch - 5ms/step
Epoch 267/300
46/46 - 0s - loss: 0.4949 - accuracy: 0.8474 - val_loss: 0.4424 - val_accuracy: 0.8837 - 245ms/epoch - 5ms/step
Epoch 268/300
46/46 - 0s - loss: 0.5031 - accuracy: 0.8502 - val_loss: 0.4748 - val_accuracy: 0.8703 - 243ms/epoch - 5ms/step
Epoch 269/300
46/46 - 0s - loss: 0.5163 - accuracy: 0.8454 - val_loss: 0.4499 - val_accuracy: 0.8753 - 241ms/epoch - 5ms/step
Epoch 270/300
46/46 - 0s - loss: 0.5025 - accuracy: 0.8464 - val_loss: 0.4253 - val_accuracy: 0.8880 - 238ms/epoch - 5ms/step
Epoch 271/300
46/46 - 0s - loss: 0.5150 - accuracy: 0.8437 - val_loss: 0.4358 - val_accuracy: 0.8833 - 243ms/epoch - 5ms/step
Epoch 272/300
46/46 - 0s - loss: 0.4986 - accuracy: 0.8439 - val_loss: 0.4468 - val_accuracy: 0.8763 - 240ms/epoch - 5ms/step
Epoch 273/300
46/46 - 0s - loss: 0.5061 - accuracy: 0.8442 - val_loss: 0.5316 - val_accuracy: 0.8560 - 238ms/epoch - 5ms/step
Epoch 274/300
46/46 - 0s - loss: 0.5195 - accuracy: 0.8393 - val_loss: 0.4803 - val_accuracy: 0.8670 - 238ms/epoch - 5ms/step
Epoch 275/300
46/46 - 0s - loss: 0.5166 - accuracy: 0.8482 - val_loss: 0.4422 - val_accuracy: 0.8830 - 240ms/epoch - 5ms/step
Epoch 276/300
46/46 - 0s - loss: 0.5174 - accuracy: 0.8395 - val_loss: 0.4662 - val_accuracy: 0.8720 - 246ms/epoch - 5ms/step
Epoch 277/300
46/46 - 0s - loss: 0.5085 - accuracy: 0.8467 - val_loss: 0.4515 - val_accuracy: 0.8813 - 252ms/epoch - 5ms/step
Epoch 278/300
46/46 - 0s - loss: 0.4958 - accuracy: 0.8476 - val_loss: 0.4559 - val_accuracy: 0.8790 - 261ms/epoch - 6ms/step
Epoch 279/300
46/46 - 0s - loss: 0.4997 - accuracy: 0.8511 - val_loss: 0.4288 - val_accuracy: 0.8847 - 259ms/epoch - 6ms/step
Epoch 280/300
46/46 - 0s - loss: 0.5179 - accuracy: 0.8456 - val_loss: 0.4428 - val_accuracy: 0.8773 - 270ms/epoch - 6ms/step
Epoch 281/300
46/46 - 0s - loss: 0.4951 - accuracy: 0.8531 - val_loss: 0.4351 - val_accuracy: 0.8837 - 254ms/epoch - 6ms/step
Epoch 282/300
46/46 - 0s - loss: 0.4945 - accuracy: 0.8560 - val_loss: 0.4848 - val_accuracy: 0.8763 - 257ms/epoch - 6ms/step
Epoch 283/300
46/46 - 0s - loss: 0.5063 - accuracy: 0.8482 - val_loss: 0.4349 - val_accuracy: 0.8843 - 256ms/epoch - 6ms/step
Epoch 284/300
46/46 - 0s - loss: 0.5034 - accuracy: 0.8458 - val_loss: 0.4469 - val_accuracy: 0.8800 - 257ms/epoch - 6ms/step
Epoch 285/300
46/46 - 0s - loss: 0.4745 - accuracy: 0.8538 - val_loss: 0.4179 - val_accuracy: 0.8890 - 253ms/epoch - 5ms/step
Epoch 286/300
46/46 - 0s - loss: 0.5006 - accuracy: 0.8482 - val_loss: 0.4341 - val_accuracy: 0.8863 - 244ms/epoch - 5ms/step
Epoch 287/300
46/46 - 0s - loss: 0.4843 - accuracy: 0.8526 - val_loss: 0.4388 - val_accuracy: 0.8800 - 242ms/epoch - 5ms/step
Epoch 288/300
46/46 - 0s - loss: 0.4917 - accuracy: 0.8488 - val_loss: 0.4437 - val_accuracy: 0.8760 - 241ms/epoch - 5ms/step
Epoch 289/300
46/46 - 0s - loss: 0.5111 - accuracy: 0.8444 - val_loss: 0.4440 - val_accuracy: 0.8793 - 240ms/epoch - 5ms/step
Epoch 290/300
46/46 - 0s - loss: 0.4873 - accuracy: 0.8528 - val_loss: 0.4143 - val_accuracy: 0.8890 - 239ms/epoch - 5ms/step
Epoch 291/300
46/46 - 0s - loss: 0.4818 - accuracy: 0.8508 - val_loss: 0.4311 - val_accuracy: 0.8803 - 239ms/epoch - 5ms/step
Epoch 292/300
46/46 - 0s - loss: 0.4751 - accuracy: 0.8523 - val_loss: 0.4321 - val_accuracy: 0.8790 - 241ms/epoch - 5ms/step
Epoch 293/300
46/46 - 0s - loss: 0.4552 - accuracy: 0.8646 - val_loss: 0.4300 - val_accuracy: 0.8857 - 239ms/epoch - 5ms/step
Epoch 294/300
46/46 - 0s - loss: 0.4781 - accuracy: 0.8559 - val_loss: 0.4335 - val_accuracy: 0.8853 - 242ms/epoch - 5ms/step
Epoch 295/300
46/46 - 0s - loss: 0.4800 - accuracy: 0.8521 - val_loss: 0.4298 - val_accuracy: 0.8867 - 237ms/epoch - 5ms/step
Epoch 296/300
46/46 - 0s - loss: 0.4866 - accuracy: 0.8519 - val_loss: 0.4510 - val_accuracy: 0.8783 - 244ms/epoch - 5ms/step
Epoch 297/300
46/46 - 0s - loss: 0.4740 - accuracy: 0.8587 - val_loss: 0.4266 - val_accuracy: 0.8840 - 239ms/epoch - 5ms/step
Epoch 298/300
46/46 - 0s - loss: 0.4726 - accuracy: 0.8529 - val_loss: 0.4145 - val_accuracy: 0.8887 - 246ms/epoch - 5ms/step
Epoch 299/300
46/46 - 0s - loss: 0.4614 - accuracy: 0.8614 - val_loss: 0.4322 - val_accuracy: 0.8820 - 258ms/epoch - 6ms/step
Epoch 300/300
46/46 - 0s - loss: 0.4753 - accuracy: 0.8541 - val_loss: 0.4827 - val_accuracy: 0.8730 - 245ms/epoch - 5ms/step
94/94 - 0s - loss: 0.4827 - accuracy: 0.8730 - 342ms/epoch - 4ms/step
Baseline Error: 12.70%

# With new data
model = GoatModel()

model.compile(optimizer='adam',
                loss= 'categorical_crossentropy',
                metrics=['accuracy'])

history = model.fit(X_train31_augmented, train_labels_augmented, validation_data=(X_val31, validation_labels),
                    epochs=500, batch_size=200, verbose=2) # no weights

scores = model.evaluate(X_val31, validation_labels, verbose=2)
print("Baseline Error: %.2f%%" % (100-scores[1]*100))
plot_history(history)
df31augmentationcompare.loc[len(df31augmentationcompare)] = ['With Augmented Engineered Data', history.history['accuracy'][-1], history.history['val_accuracy'][-1], history.history['loss'][-1], history.history['val_loss'][-1], history]

Epoch 1/500

2023-11-26 13:27:09.474091: E tensorflow/core/grappler/optimizers/meta_optimizer.cc:954] layout failed: INVALID_ARGUMENT: Size of values 0 does not match size of permutation 4 @ fanin shape insequential_13/dropout_15/dropout/SelectV2-2-TransposeNHWCToNCHW-LayoutOptimizer

72/72 - 3s - loss: 2.7526 - accuracy: 0.0869 - val_loss: 2.5949 - val_accuracy: 0.1367 - 3s/epoch - 36ms/step
Epoch 2/500
72/72 - 0s - loss: 2.5484 - accuracy: 0.1427 - val_loss: 2.4543 - val_accuracy: 0.2050 - 357ms/epoch - 5ms/step
Epoch 3/500
72/72 - 0s - loss: 2.3749 - accuracy: 0.2001 - val_loss: 2.3098 - val_accuracy: 0.2493 - 341ms/epoch - 5ms/step
Epoch 4/500
72/72 - 0s - loss: 2.2701 - accuracy: 0.2444 - val_loss: 2.1625 - val_accuracy: 0.3007 - 341ms/epoch - 5ms/step
Epoch 5/500
72/72 - 0s - loss: 2.1884 - accuracy: 0.2830 - val_loss: 1.9654 - val_accuracy: 0.4050 - 339ms/epoch - 5ms/step
Epoch 6/500
72/72 - 0s - loss: 2.0979 - accuracy: 0.3150 - val_loss: 1.9631 - val_accuracy: 0.3833 - 337ms/epoch - 5ms/step
Epoch 7/500
72/72 - 0s - loss: 2.0311 - accuracy: 0.3419 - val_loss: 1.8587 - val_accuracy: 0.4070 - 335ms/epoch - 5ms/step
Epoch 8/500
72/72 - 0s - loss: 1.9872 - accuracy: 0.3608 - val_loss: 1.7038 - val_accuracy: 0.4857 - 339ms/epoch - 5ms/step
Epoch 9/500
72/72 - 0s - loss: 1.9299 - accuracy: 0.3773 - val_loss: 1.6106 - val_accuracy: 0.5230 - 334ms/epoch - 5ms/step
Epoch 10/500
72/72 - 0s - loss: 1.8679 - accuracy: 0.3932 - val_loss: 1.6287 - val_accuracy: 0.5083 - 336ms/epoch - 5ms/step
Epoch 11/500
72/72 - 0s - loss: 1.8507 - accuracy: 0.4044 - val_loss: 1.5495 - val_accuracy: 0.5313 - 341ms/epoch - 5ms/step
Epoch 12/500
72/72 - 0s - loss: 1.7969 - accuracy: 0.4236 - val_loss: 1.5041 - val_accuracy: 0.5533 - 338ms/epoch - 5ms/step
Epoch 13/500
72/72 - 0s - loss: 1.7810 - accuracy: 0.4275 - val_loss: 1.4944 - val_accuracy: 0.5530 - 336ms/epoch - 5ms/step
Epoch 14/500
72/72 - 0s - loss: 1.7604 - accuracy: 0.4408 - val_loss: 1.5612 - val_accuracy: 0.5063 - 342ms/epoch - 5ms/step
Epoch 15/500
72/72 - 0s - loss: 1.7127 - accuracy: 0.4543 - val_loss: 1.4232 - val_accuracy: 0.5770 - 339ms/epoch - 5ms/step
Epoch 16/500
72/72 - 0s - loss: 1.6956 - accuracy: 0.4577 - val_loss: 1.4221 - val_accuracy: 0.5753 - 339ms/epoch - 5ms/step
Epoch 17/500
72/72 - 0s - loss: 1.6708 - accuracy: 0.4662 - val_loss: 1.3078 - val_accuracy: 0.6227 - 339ms/epoch - 5ms/step
Epoch 18/500
72/72 - 0s - loss: 1.6515 - accuracy: 0.4766 - val_loss: 1.2686 - val_accuracy: 0.6237 - 334ms/epoch - 5ms/step
Epoch 19/500
72/72 - 0s - loss: 1.6281 - accuracy: 0.4782 - val_loss: 1.3858 - val_accuracy: 0.5713 - 337ms/epoch - 5ms/step
Epoch 20/500
72/72 - 0s - loss: 1.6186 - accuracy: 0.4902 - val_loss: 1.2254 - val_accuracy: 0.6433 - 336ms/epoch - 5ms/step
Epoch 21/500
72/72 - 0s - loss: 1.5934 - accuracy: 0.4983 - val_loss: 1.2732 - val_accuracy: 0.6247 - 337ms/epoch - 5ms/step
Epoch 22/500
72/72 - 0s - loss: 1.5892 - accuracy: 0.4982 - val_loss: 1.1872 - val_accuracy: 0.6697 - 338ms/epoch - 5ms/step
Epoch 23/500
72/72 - 0s - loss: 1.5503 - accuracy: 0.5056 - val_loss: 1.1484 - val_accuracy: 0.6670 - 335ms/epoch - 5ms/step
Epoch 24/500
72/72 - 0s - loss: 1.5402 - accuracy: 0.5162 - val_loss: 1.1565 - val_accuracy: 0.6660 - 335ms/epoch - 5ms/step
Epoch 25/500
72/72 - 0s - loss: 1.5279 - accuracy: 0.5201 - val_loss: 1.1275 - val_accuracy: 0.6723 - 334ms/epoch - 5ms/step
Epoch 26/500
72/72 - 0s - loss: 1.5153 - accuracy: 0.5280 - val_loss: 1.1924 - val_accuracy: 0.6517 - 331ms/epoch - 5ms/step
Epoch 27/500
72/72 - 0s - loss: 1.4798 - accuracy: 0.5334 - val_loss: 1.1424 - val_accuracy: 0.6763 - 331ms/epoch - 5ms/step
Epoch 28/500
72/72 - 0s - loss: 1.4708 - accuracy: 0.5392 - val_loss: 1.1081 - val_accuracy: 0.6820 - 332ms/epoch - 5ms/step
Epoch 29/500
72/72 - 0s - loss: 1.4690 - accuracy: 0.5458 - val_loss: 1.0764 - val_accuracy: 0.6983 - 347ms/epoch - 5ms/step
Epoch 30/500
72/72 - 0s - loss: 1.4387 - accuracy: 0.5498 - val_loss: 1.0471 - val_accuracy: 0.6930 - 345ms/epoch - 5ms/step
Epoch 31/500
72/72 - 0s - loss: 1.4280 - accuracy: 0.5560 - val_loss: 1.0340 - val_accuracy: 0.7073 - 342ms/epoch - 5ms/step
Epoch 32/500
72/72 - 0s - loss: 1.4296 - accuracy: 0.5500 - val_loss: 1.0224 - val_accuracy: 0.7160 - 335ms/epoch - 5ms/step
Epoch 33/500
72/72 - 0s - loss: 1.3958 - accuracy: 0.5594 - val_loss: 1.0492 - val_accuracy: 0.6960 - 334ms/epoch - 5ms/step
Epoch 34/500
72/72 - 0s - loss: 1.3888 - accuracy: 0.5668 - val_loss: 0.9689 - val_accuracy: 0.7217 - 336ms/epoch - 5ms/step
Epoch 35/500
72/72 - 0s - loss: 1.3759 - accuracy: 0.5654 - val_loss: 1.0518 - val_accuracy: 0.6943 - 330ms/epoch - 5ms/step
Epoch 36/500
72/72 - 0s - loss: 1.3471 - accuracy: 0.5830 - val_loss: 0.9737 - val_accuracy: 0.7250 - 328ms/epoch - 5ms/step
Epoch 37/500
72/72 - 0s - loss: 1.3433 - accuracy: 0.5807 - val_loss: 0.9128 - val_accuracy: 0.7333 - 334ms/epoch - 5ms/step
Epoch 38/500
72/72 - 0s - loss: 1.3443 - accuracy: 0.5825 - val_loss: 0.9332 - val_accuracy: 0.7303 - 348ms/epoch - 5ms/step
Epoch 39/500
72/72 - 0s - loss: 1.3490 - accuracy: 0.5794 - val_loss: 0.9190 - val_accuracy: 0.7317 - 343ms/epoch - 5ms/step
Epoch 40/500
72/72 - 0s - loss: 1.3337 - accuracy: 0.5838 - val_loss: 0.9319 - val_accuracy: 0.7283 - 339ms/epoch - 5ms/step
Epoch 41/500
72/72 - 0s - loss: 1.3155 - accuracy: 0.5934 - val_loss: 0.9677 - val_accuracy: 0.7113 - 331ms/epoch - 5ms/step
Epoch 42/500
72/72 - 0s - loss: 1.3167 - accuracy: 0.5888 - val_loss: 0.9005 - val_accuracy: 0.7473 - 332ms/epoch - 5ms/step
Epoch 43/500
72/72 - 0s - loss: 1.2987 - accuracy: 0.5934 - val_loss: 0.9090 - val_accuracy: 0.7430 - 333ms/epoch - 5ms/step
Epoch 44/500
72/72 - 0s - loss: 1.2850 - accuracy: 0.5984 - val_loss: 0.9623 - val_accuracy: 0.7170 - 334ms/epoch - 5ms/step
Epoch 45/500
72/72 - 0s - loss: 1.2722 - accuracy: 0.6036 - val_loss: 0.9047 - val_accuracy: 0.7393 - 330ms/epoch - 5ms/step
Epoch 46/500
72/72 - 0s - loss: 1.2696 - accuracy: 0.6082 - val_loss: 0.8829 - val_accuracy: 0.7377 - 333ms/epoch - 5ms/step
Epoch 47/500
72/72 - 0s - loss: 1.2634 - accuracy: 0.6036 - val_loss: 0.9058 - val_accuracy: 0.7433 - 343ms/epoch - 5ms/step
Epoch 48/500
72/72 - 0s - loss: 1.2493 - accuracy: 0.6133 - val_loss: 0.8826 - val_accuracy: 0.7487 - 339ms/epoch - 5ms/step
Epoch 49/500
72/72 - 0s - loss: 1.2428 - accuracy: 0.6150 - val_loss: 0.8381 - val_accuracy: 0.7640 - 337ms/epoch - 5ms/step
Epoch 50/500
72/72 - 0s - loss: 1.2365 - accuracy: 0.6177 - val_loss: 0.8315 - val_accuracy: 0.7640 - 341ms/epoch - 5ms/step
Epoch 51/500
72/72 - 0s - loss: 1.2301 - accuracy: 0.6165 - val_loss: 0.8311 - val_accuracy: 0.7627 - 335ms/epoch - 5ms/step
Epoch 52/500
72/72 - 0s - loss: 1.2219 - accuracy: 0.6222 - val_loss: 0.8681 - val_accuracy: 0.7563 - 335ms/epoch - 5ms/step
Epoch 53/500
72/72 - 0s - loss: 1.2106 - accuracy: 0.6243 - val_loss: 0.8480 - val_accuracy: 0.7570 - 332ms/epoch - 5ms/step
Epoch 54/500
72/72 - 0s - loss: 1.2027 - accuracy: 0.6252 - val_loss: 0.8714 - val_accuracy: 0.7453 - 338ms/epoch - 5ms/step
Epoch 55/500
72/72 - 0s - loss: 1.2061 - accuracy: 0.6248 - val_loss: 0.7984 - val_accuracy: 0.7763 - 339ms/epoch - 5ms/step
Epoch 56/500
72/72 - 0s - loss: 1.1989 - accuracy: 0.6290 - val_loss: 0.7952 - val_accuracy: 0.7813 - 335ms/epoch - 5ms/step
Epoch 57/500
72/72 - 0s - loss: 1.1782 - accuracy: 0.6397 - val_loss: 0.7553 - val_accuracy: 0.7900 - 334ms/epoch - 5ms/step
Epoch 58/500
72/72 - 0s - loss: 1.1692 - accuracy: 0.6399 - val_loss: 0.7816 - val_accuracy: 0.7723 - 343ms/epoch - 5ms/step
Epoch 59/500
72/72 - 0s - loss: 1.1726 - accuracy: 0.6386 - val_loss: 0.7882 - val_accuracy: 0.7800 - 346ms/epoch - 5ms/step
Epoch 60/500
72/72 - 0s - loss: 1.1643 - accuracy: 0.6372 - val_loss: 0.8063 - val_accuracy: 0.7710 - 335ms/epoch - 5ms/step
Epoch 61/500
72/72 - 0s - loss: 1.1629 - accuracy: 0.6413 - val_loss: 0.7470 - val_accuracy: 0.7837 - 336ms/epoch - 5ms/step
Epoch 62/500
72/72 - 0s - loss: 1.1476 - accuracy: 0.6452 - val_loss: 0.7311 - val_accuracy: 0.7940 - 340ms/epoch - 5ms/step
Epoch 63/500
72/72 - 0s - loss: 1.1413 - accuracy: 0.6447 - val_loss: 0.7331 - val_accuracy: 0.7913 - 341ms/epoch - 5ms/step
Epoch 64/500
72/72 - 0s - loss: 1.1287 - accuracy: 0.6497 - val_loss: 0.7604 - val_accuracy: 0.7853 - 340ms/epoch - 5ms/step
Epoch 65/500
72/72 - 0s - loss: 1.1321 - accuracy: 0.6484 - val_loss: 0.7535 - val_accuracy: 0.7833 - 340ms/epoch - 5ms/step
Epoch 66/500
72/72 - 0s - loss: 1.1339 - accuracy: 0.6507 - val_loss: 0.7170 - val_accuracy: 0.7893 - 346ms/epoch - 5ms/step
Epoch 67/500
72/72 - 0s - loss: 1.1281 - accuracy: 0.6530 - val_loss: 0.7566 - val_accuracy: 0.7873 - 348ms/epoch - 5ms/step
Epoch 68/500
72/72 - 0s - loss: 1.1110 - accuracy: 0.6602 - val_loss: 0.6820 - val_accuracy: 0.8150 - 342ms/epoch - 5ms/step
Epoch 69/500
72/72 - 0s - loss: 1.1003 - accuracy: 0.6588 - val_loss: 0.7531 - val_accuracy: 0.7863 - 336ms/epoch - 5ms/step
Epoch 70/500
72/72 - 0s - loss: 1.1038 - accuracy: 0.6567 - val_loss: 0.7339 - val_accuracy: 0.7953 - 340ms/epoch - 5ms/step
Epoch 71/500
72/72 - 0s - loss: 1.0872 - accuracy: 0.6628 - val_loss: 0.7389 - val_accuracy: 0.7877 - 343ms/epoch - 5ms/step
Epoch 72/500
72/72 - 0s - loss: 1.0846 - accuracy: 0.6659 - val_loss: 0.7366 - val_accuracy: 0.7873 - 337ms/epoch - 5ms/step
Epoch 73/500
72/72 - 0s - loss: 1.0864 - accuracy: 0.6604 - val_loss: 0.6824 - val_accuracy: 0.8083 - 334ms/epoch - 5ms/step
Epoch 74/500
72/72 - 0s - loss: 1.0821 - accuracy: 0.6640 - val_loss: 0.6918 - val_accuracy: 0.8097 - 338ms/epoch - 5ms/step
Epoch 75/500
72/72 - 0s - loss: 1.0723 - accuracy: 0.6661 - val_loss: 0.7224 - val_accuracy: 0.7920 - 339ms/epoch - 5ms/step
Epoch 76/500
72/72 - 0s - loss: 1.0836 - accuracy: 0.6654 - val_loss: 0.6751 - val_accuracy: 0.8107 - 339ms/epoch - 5ms/step
Epoch 77/500
72/72 - 0s - loss: 1.0665 - accuracy: 0.6754 - val_loss: 0.6731 - val_accuracy: 0.8170 - 340ms/epoch - 5ms/step
Epoch 78/500
72/72 - 0s - loss: 1.0588 - accuracy: 0.6763 - val_loss: 0.6638 - val_accuracy: 0.8143 - 338ms/epoch - 5ms/step
Epoch 79/500
72/72 - 0s - loss: 1.0461 - accuracy: 0.6748 - val_loss: 0.6849 - val_accuracy: 0.8020 - 337ms/epoch - 5ms/step
Epoch 80/500
72/72 - 0s - loss: 1.0473 - accuracy: 0.6766 - val_loss: 0.6619 - val_accuracy: 0.8103 - 340ms/epoch - 5ms/step
Epoch 81/500
72/72 - 0s - loss: 1.0431 - accuracy: 0.6744 - val_loss: 0.7143 - val_accuracy: 0.7907 - 337ms/epoch - 5ms/step
Epoch 82/500
72/72 - 0s - loss: 1.0439 - accuracy: 0.6768 - val_loss: 0.6507 - val_accuracy: 0.8170 - 342ms/epoch - 5ms/step
Epoch 83/500
72/72 - 0s - loss: 1.0243 - accuracy: 0.6799 - val_loss: 0.6400 - val_accuracy: 0.8170 - 335ms/epoch - 5ms/step
Epoch 84/500
72/72 - 0s - loss: 1.0476 - accuracy: 0.6813 - val_loss: 0.8161 - val_accuracy: 0.7650 - 340ms/epoch - 5ms/step
Epoch 85/500
72/72 - 0s - loss: 1.0362 - accuracy: 0.6788 - val_loss: 0.6650 - val_accuracy: 0.8103 - 337ms/epoch - 5ms/step
Epoch 86/500
72/72 - 0s - loss: 1.0130 - accuracy: 0.6880 - val_loss: 0.6254 - val_accuracy: 0.8193 - 337ms/epoch - 5ms/step
Epoch 87/500
72/72 - 0s - loss: 1.0210 - accuracy: 0.6891 - val_loss: 0.6137 - val_accuracy: 0.8277 - 337ms/epoch - 5ms/step
Epoch 88/500
72/72 - 0s - loss: 1.0164 - accuracy: 0.6877 - val_loss: 0.6086 - val_accuracy: 0.8323 - 338ms/epoch - 5ms/step
Epoch 89/500
72/72 - 0s - loss: 0.9990 - accuracy: 0.6873 - val_loss: 0.6015 - val_accuracy: 0.8247 - 334ms/epoch - 5ms/step
Epoch 90/500
72/72 - 0s - loss: 1.0065 - accuracy: 0.6931 - val_loss: 0.6650 - val_accuracy: 0.8087 - 336ms/epoch - 5ms/step
Epoch 91/500
72/72 - 0s - loss: 0.9843 - accuracy: 0.6963 - val_loss: 0.6198 - val_accuracy: 0.8233 - 336ms/epoch - 5ms/step
Epoch 92/500
72/72 - 0s - loss: 1.0065 - accuracy: 0.6956 - val_loss: 0.5978 - val_accuracy: 0.8313 - 344ms/epoch - 5ms/step
Epoch 93/500
72/72 - 0s - loss: 0.9878 - accuracy: 0.6977 - val_loss: 0.6104 - val_accuracy: 0.8313 - 339ms/epoch - 5ms/step
Epoch 94/500
72/72 - 0s - loss: 0.9800 - accuracy: 0.7002 - val_loss: 0.6061 - val_accuracy: 0.8303 - 332ms/epoch - 5ms/step
Epoch 95/500
72/72 - 0s - loss: 0.9908 - accuracy: 0.6961 - val_loss: 0.6002 - val_accuracy: 0.8210 - 332ms/epoch - 5ms/step
Epoch 96/500
72/72 - 0s - loss: 0.9845 - accuracy: 0.6985 - val_loss: 0.6078 - val_accuracy: 0.8243 - 335ms/epoch - 5ms/step
Epoch 97/500
72/72 - 0s - loss: 0.9539 - accuracy: 0.7099 - val_loss: 0.6114 - val_accuracy: 0.8213 - 334ms/epoch - 5ms/step
Epoch 98/500
72/72 - 0s - loss: 0.9698 - accuracy: 0.7039 - val_loss: 0.6155 - val_accuracy: 0.8250 - 335ms/epoch - 5ms/step
Epoch 99/500
72/72 - 0s - loss: 0.9690 - accuracy: 0.6996 - val_loss: 0.6002 - val_accuracy: 0.8267 - 340ms/epoch - 5ms/step
Epoch 100/500
72/72 - 0s - loss: 0.9628 - accuracy: 0.7030 - val_loss: 0.6210 - val_accuracy: 0.8227 - 338ms/epoch - 5ms/step
Epoch 101/500
72/72 - 0s - loss: 0.9559 - accuracy: 0.7075 - val_loss: 0.5900 - val_accuracy: 0.8320 - 339ms/epoch - 5ms/step
Epoch 102/500
72/72 - 0s - loss: 0.9557 - accuracy: 0.7068 - val_loss: 0.5938 - val_accuracy: 0.8300 - 335ms/epoch - 5ms/step
Epoch 103/500
72/72 - 0s - loss: 0.9495 - accuracy: 0.7149 - val_loss: 0.6059 - val_accuracy: 0.8300 - 335ms/epoch - 5ms/step
Epoch 104/500
72/72 - 0s - loss: 0.9471 - accuracy: 0.7109 - val_loss: 0.5730 - val_accuracy: 0.8393 - 337ms/epoch - 5ms/step
Epoch 105/500
72/72 - 0s - loss: 0.9455 - accuracy: 0.7113 - val_loss: 0.5759 - val_accuracy: 0.8360 - 340ms/epoch - 5ms/step
Epoch 106/500
72/72 - 0s - loss: 0.9455 - accuracy: 0.7108 - val_loss: 0.6316 - val_accuracy: 0.8143 - 345ms/epoch - 5ms/step
Epoch 107/500
72/72 - 0s - loss: 0.9334 - accuracy: 0.7110 - val_loss: 0.5942 - val_accuracy: 0.8303 - 335ms/epoch - 5ms/step
Epoch 108/500
72/72 - 0s - loss: 0.9313 - accuracy: 0.7143 - val_loss: 0.5608 - val_accuracy: 0.8370 - 337ms/epoch - 5ms/step
Epoch 109/500
72/72 - 0s - loss: 0.9241 - accuracy: 0.7181 - val_loss: 0.6094 - val_accuracy: 0.8200 - 339ms/epoch - 5ms/step
Epoch 110/500
72/72 - 0s - loss: 0.9245 - accuracy: 0.7190 - val_loss: 0.5999 - val_accuracy: 0.8230 - 346ms/epoch - 5ms/step
Epoch 111/500
72/72 - 0s - loss: 0.9192 - accuracy: 0.7198 - val_loss: 0.5337 - val_accuracy: 0.8503 - 334ms/epoch - 5ms/step
Epoch 112/500
72/72 - 0s - loss: 0.9203 - accuracy: 0.7155 - val_loss: 0.5610 - val_accuracy: 0.8370 - 336ms/epoch - 5ms/step
Epoch 113/500
72/72 - 0s - loss: 0.9166 - accuracy: 0.7188 - val_loss: 0.5393 - val_accuracy: 0.8477 - 337ms/epoch - 5ms/step
Epoch 114/500
72/72 - 0s - loss: 0.9284 - accuracy: 0.7122 - val_loss: 0.6185 - val_accuracy: 0.8207 - 338ms/epoch - 5ms/step
Epoch 115/500
72/72 - 0s - loss: 0.9115 - accuracy: 0.7197 - val_loss: 0.5296 - val_accuracy: 0.8500 - 339ms/epoch - 5ms/step
Epoch 116/500
72/72 - 0s - loss: 0.9105 - accuracy: 0.7208 - val_loss: 0.5362 - val_accuracy: 0.8490 - 336ms/epoch - 5ms/step
Epoch 117/500
72/72 - 0s - loss: 0.9025 - accuracy: 0.7240 - val_loss: 0.5185 - val_accuracy: 0.8510 - 346ms/epoch - 5ms/step
Epoch 118/500
72/72 - 0s - loss: 0.9111 - accuracy: 0.7170 - val_loss: 0.5648 - val_accuracy: 0.8393 - 347ms/epoch - 5ms/step
Epoch 119/500
72/72 - 0s - loss: 0.8842 - accuracy: 0.7273 - val_loss: 0.5837 - val_accuracy: 0.8310 - 342ms/epoch - 5ms/step
Epoch 120/500
72/72 - 0s - loss: 0.8891 - accuracy: 0.7283 - val_loss: 0.5885 - val_accuracy: 0.8277 - 336ms/epoch - 5ms/step
Epoch 121/500
72/72 - 0s - loss: 0.9000 - accuracy: 0.7252 - val_loss: 0.5280 - val_accuracy: 0.8497 - 332ms/epoch - 5ms/step
Epoch 122/500
72/72 - 0s - loss: 0.8901 - accuracy: 0.7299 - val_loss: 0.5315 - val_accuracy: 0.8513 - 344ms/epoch - 5ms/step
Epoch 123/500
72/72 - 0s - loss: 0.8729 - accuracy: 0.7346 - val_loss: 0.5241 - val_accuracy: 0.8493 - 339ms/epoch - 5ms/step
Epoch 124/500
72/72 - 0s - loss: 0.8707 - accuracy: 0.7351 - val_loss: 0.5344 - val_accuracy: 0.8500 - 336ms/epoch - 5ms/step
Epoch 125/500
72/72 - 0s - loss: 0.8752 - accuracy: 0.7289 - val_loss: 0.5197 - val_accuracy: 0.8530 - 337ms/epoch - 5ms/step
Epoch 126/500
72/72 - 0s - loss: 0.8708 - accuracy: 0.7356 - val_loss: 0.5151 - val_accuracy: 0.8547 - 343ms/epoch - 5ms/step
Epoch 127/500
72/72 - 0s - loss: 0.8747 - accuracy: 0.7331 - val_loss: 0.5324 - val_accuracy: 0.8513 - 336ms/epoch - 5ms/step
Epoch 128/500
72/72 - 0s - loss: 0.8727 - accuracy: 0.7343 - val_loss: 0.5243 - val_accuracy: 0.8543 - 340ms/epoch - 5ms/step
Epoch 129/500
72/72 - 0s - loss: 0.8558 - accuracy: 0.7326 - val_loss: 0.5057 - val_accuracy: 0.8563 - 352ms/epoch - 5ms/step
Epoch 130/500
72/72 - 0s - loss: 0.8689 - accuracy: 0.7355 - val_loss: 0.5791 - val_accuracy: 0.8347 - 347ms/epoch - 5ms/step
Epoch 131/500
72/72 - 0s - loss: 0.8597 - accuracy: 0.7401 - val_loss: 0.4949 - val_accuracy: 0.8593 - 338ms/epoch - 5ms/step
Epoch 132/500
72/72 - 0s - loss: 0.8559 - accuracy: 0.7377 - val_loss: 0.4985 - val_accuracy: 0.8657 - 331ms/epoch - 5ms/step
Epoch 133/500
72/72 - 0s - loss: 0.8578 - accuracy: 0.7331 - val_loss: 0.5504 - val_accuracy: 0.8447 - 331ms/epoch - 5ms/step
Epoch 134/500
72/72 - 0s - loss: 0.8542 - accuracy: 0.7393 - val_loss: 0.5193 - val_accuracy: 0.8597 - 336ms/epoch - 5ms/step
Epoch 135/500
72/72 - 0s - loss: 0.8505 - accuracy: 0.7388 - val_loss: 0.5329 - val_accuracy: 0.8453 - 333ms/epoch - 5ms/step
Epoch 136/500
72/72 - 0s - loss: 0.8426 - accuracy: 0.7439 - val_loss: 0.5255 - val_accuracy: 0.8507 - 338ms/epoch - 5ms/step
Epoch 137/500
72/72 - 0s - loss: 0.8410 - accuracy: 0.7409 - val_loss: 0.4943 - val_accuracy: 0.8560 - 338ms/epoch - 5ms/step
Epoch 138/500
72/72 - 0s - loss: 0.8361 - accuracy: 0.7421 - val_loss: 0.5016 - val_accuracy: 0.8613 - 334ms/epoch - 5ms/step
Epoch 139/500
72/72 - 0s - loss: 0.8370 - accuracy: 0.7478 - val_loss: 0.4941 - val_accuracy: 0.8667 - 334ms/epoch - 5ms/step
Epoch 140/500
72/72 - 0s - loss: 0.8358 - accuracy: 0.7435 - val_loss: 0.4815 - val_accuracy: 0.8633 - 338ms/epoch - 5ms/step
Epoch 141/500
72/72 - 0s - loss: 0.8219 - accuracy: 0.7471 - val_loss: 0.5473 - val_accuracy: 0.8457 - 335ms/epoch - 5ms/step
Epoch 142/500
72/72 - 0s - loss: 0.8308 - accuracy: 0.7496 - val_loss: 0.4887 - val_accuracy: 0.8627 - 336ms/epoch - 5ms/step
Epoch 143/500
72/72 - 0s - loss: 0.8243 - accuracy: 0.7457 - val_loss: 0.5089 - val_accuracy: 0.8553 - 338ms/epoch - 5ms/step
Epoch 144/500
72/72 - 0s - loss: 0.8294 - accuracy: 0.7463 - val_loss: 0.5139 - val_accuracy: 0.8530 - 338ms/epoch - 5ms/step
Epoch 145/500
72/72 - 0s - loss: 0.8193 - accuracy: 0.7487 - val_loss: 0.5448 - val_accuracy: 0.8410 - 333ms/epoch - 5ms/step
Epoch 146/500
72/72 - 0s - loss: 0.8117 - accuracy: 0.7535 - val_loss: 0.5226 - val_accuracy: 0.8513 - 339ms/epoch - 5ms/step
Epoch 147/500
72/72 - 0s - loss: 0.8165 - accuracy: 0.7509 - val_loss: 0.4780 - val_accuracy: 0.8697 - 338ms/epoch - 5ms/step
Epoch 148/500
72/72 - 0s - loss: 0.8050 - accuracy: 0.7558 - val_loss: 0.4989 - val_accuracy: 0.8550 - 341ms/epoch - 5ms/step
Epoch 149/500
72/72 - 0s - loss: 0.8169 - accuracy: 0.7524 - val_loss: 0.4781 - val_accuracy: 0.8620 - 340ms/epoch - 5ms/step
Epoch 150/500
72/72 - 0s - loss: 0.8166 - accuracy: 0.7465 - val_loss: 0.4865 - val_accuracy: 0.8650 - 359ms/epoch - 5ms/step
Epoch 151/500
72/72 - 0s - loss: 0.8081 - accuracy: 0.7518 - val_loss: 0.4765 - val_accuracy: 0.8667 - 358ms/epoch - 5ms/step
Epoch 152/500
72/72 - 0s - loss: 0.8130 - accuracy: 0.7472 - val_loss: 0.4913 - val_accuracy: 0.8640 - 393ms/epoch - 5ms/step
Epoch 153/500
72/72 - 0s - loss: 0.8107 - accuracy: 0.7526 - val_loss: 0.4713 - val_accuracy: 0.8707 - 377ms/epoch - 5ms/step
Epoch 154/500
72/72 - 0s - loss: 0.8099 - accuracy: 0.7525 - val_loss: 0.4643 - val_accuracy: 0.8733 - 389ms/epoch - 5ms/step
Epoch 155/500
72/72 - 0s - loss: 0.7947 - accuracy: 0.7623 - val_loss: 0.4778 - val_accuracy: 0.8657 - 375ms/epoch - 5ms/step
Epoch 156/500
72/72 - 0s - loss: 0.7978 - accuracy: 0.7565 - val_loss: 0.4876 - val_accuracy: 0.8620 - 361ms/epoch - 5ms/step
Epoch 157/500
72/72 - 0s - loss: 0.7932 - accuracy: 0.7594 - val_loss: 0.4896 - val_accuracy: 0.8607 - 340ms/epoch - 5ms/step
Epoch 158/500
72/72 - 0s - loss: 0.8044 - accuracy: 0.7556 - val_loss: 0.4685 - val_accuracy: 0.8690 - 334ms/epoch - 5ms/step
Epoch 159/500
72/72 - 0s - loss: 0.7851 - accuracy: 0.7647 - val_loss: 0.4619 - val_accuracy: 0.8653 - 336ms/epoch - 5ms/step
Epoch 160/500
72/72 - 0s - loss: 0.7921 - accuracy: 0.7550 - val_loss: 0.4522 - val_accuracy: 0.8747 - 339ms/epoch - 5ms/step
Epoch 161/500
72/72 - 0s - loss: 0.7845 - accuracy: 0.7638 - val_loss: 0.5084 - val_accuracy: 0.8547 - 330ms/epoch - 5ms/step
Epoch 162/500
72/72 - 0s - loss: 0.7923 - accuracy: 0.7585 - val_loss: 0.4751 - val_accuracy: 0.8670 - 333ms/epoch - 5ms/step
Epoch 163/500
72/72 - 0s - loss: 0.7756 - accuracy: 0.7611 - val_loss: 0.4725 - val_accuracy: 0.8673 - 332ms/epoch - 5ms/step
Epoch 164/500
72/72 - 0s - loss: 0.7628 - accuracy: 0.7681 - val_loss: 0.4493 - val_accuracy: 0.8733 - 334ms/epoch - 5ms/step
Epoch 165/500
72/72 - 0s - loss: 0.7700 - accuracy: 0.7678 - val_loss: 0.4716 - val_accuracy: 0.8657 - 336ms/epoch - 5ms/step
Epoch 166/500
72/72 - 0s - loss: 0.7761 - accuracy: 0.7644 - val_loss: 0.4630 - val_accuracy: 0.8720 - 336ms/epoch - 5ms/step
Epoch 167/500
72/72 - 0s - loss: 0.7759 - accuracy: 0.7606 - val_loss: 0.4977 - val_accuracy: 0.8580 - 340ms/epoch - 5ms/step
Epoch 168/500
72/72 - 0s - loss: 0.7722 - accuracy: 0.7657 - val_loss: 0.4526 - val_accuracy: 0.8733 - 338ms/epoch - 5ms/step
Epoch 169/500
72/72 - 0s - loss: 0.7773 - accuracy: 0.7642 - val_loss: 0.5254 - val_accuracy: 0.8463 - 336ms/epoch - 5ms/step
Epoch 170/500
72/72 - 0s - loss: 0.7909 - accuracy: 0.7619 - val_loss: 0.4545 - val_accuracy: 0.8710 - 343ms/epoch - 5ms/step
Epoch 171/500
72/72 - 0s - loss: 0.7695 - accuracy: 0.7650 - val_loss: 0.4569 - val_accuracy: 0.8740 - 335ms/epoch - 5ms/step
Epoch 172/500
72/72 - 0s - loss: 0.7680 - accuracy: 0.7640 - val_loss: 0.4779 - val_accuracy: 0.8607 - 336ms/epoch - 5ms/step
Epoch 173/500
72/72 - 0s - loss: 0.7711 - accuracy: 0.7636 - val_loss: 0.4800 - val_accuracy: 0.8650 - 336ms/epoch - 5ms/step
Epoch 174/500
72/72 - 0s - loss: 0.7552 - accuracy: 0.7664 - val_loss: 0.4535 - val_accuracy: 0.8707 - 337ms/epoch - 5ms/step
Epoch 175/500
72/72 - 0s - loss: 0.7497 - accuracy: 0.7724 - val_loss: 0.4458 - val_accuracy: 0.8757 - 332ms/epoch - 5ms/step
Epoch 176/500
72/72 - 0s - loss: 0.7511 - accuracy: 0.7721 - val_loss: 0.4450 - val_accuracy: 0.8780 - 334ms/epoch - 5ms/step
Epoch 177/500
72/72 - 0s - loss: 0.7595 - accuracy: 0.7673 - val_loss: 0.4529 - val_accuracy: 0.8720 - 335ms/epoch - 5ms/step
Epoch 178/500
72/72 - 0s - loss: 0.7588 - accuracy: 0.7733 - val_loss: 0.4469 - val_accuracy: 0.8783 - 332ms/epoch - 5ms/step
Epoch 179/500
72/72 - 0s - loss: 0.7622 - accuracy: 0.7696 - val_loss: 0.4834 - val_accuracy: 0.8647 - 334ms/epoch - 5ms/step
Epoch 180/500
72/72 - 0s - loss: 0.7517 - accuracy: 0.7732 - val_loss: 0.4876 - val_accuracy: 0.8633 - 337ms/epoch - 5ms/step
Epoch 181/500
72/72 - 0s - loss: 0.7626 - accuracy: 0.7668 - val_loss: 0.4641 - val_accuracy: 0.8700 - 332ms/epoch - 5ms/step
Epoch 182/500
72/72 - 0s - loss: 0.7504 - accuracy: 0.7680 - val_loss: 0.4561 - val_accuracy: 0.8723 - 335ms/epoch - 5ms/step
Epoch 183/500
72/72 - 0s - loss: 0.7439 - accuracy: 0.7733 - val_loss: 0.4407 - val_accuracy: 0.8750 - 338ms/epoch - 5ms/step
Epoch 184/500
72/72 - 0s - loss: 0.7495 - accuracy: 0.7692 - val_loss: 0.4405 - val_accuracy: 0.8740 - 333ms/epoch - 5ms/step
Epoch 185/500
72/72 - 0s - loss: 0.7512 - accuracy: 0.7719 - val_loss: 0.4780 - val_accuracy: 0.8680 - 332ms/epoch - 5ms/step
Epoch 186/500
72/72 - 0s - loss: 0.7370 - accuracy: 0.7766 - val_loss: 0.4403 - val_accuracy: 0.8737 - 339ms/epoch - 5ms/step
Epoch 187/500
72/72 - 0s - loss: 0.7371 - accuracy: 0.7771 - val_loss: 0.4515 - val_accuracy: 0.8677 - 333ms/epoch - 5ms/step
Epoch 188/500
72/72 - 0s - loss: 0.7429 - accuracy: 0.7747 - val_loss: 0.4279 - val_accuracy: 0.8803 - 335ms/epoch - 5ms/step
Epoch 189/500
72/72 - 0s - loss: 0.7308 - accuracy: 0.7752 - val_loss: 0.4308 - val_accuracy: 0.8807 - 335ms/epoch - 5ms/step
Epoch 190/500
72/72 - 0s - loss: 0.7228 - accuracy: 0.7764 - val_loss: 0.4494 - val_accuracy: 0.8680 - 333ms/epoch - 5ms/step
Epoch 191/500
72/72 - 0s - loss: 0.7417 - accuracy: 0.7747 - val_loss: 0.4354 - val_accuracy: 0.8760 - 333ms/epoch - 5ms/step
Epoch 192/500
72/72 - 0s - loss: 0.7471 - accuracy: 0.7735 - val_loss: 0.4620 - val_accuracy: 0.8683 - 334ms/epoch - 5ms/step
Epoch 193/500
72/72 - 0s - loss: 0.7209 - accuracy: 0.7795 - val_loss: 0.4166 - val_accuracy: 0.8807 - 336ms/epoch - 5ms/step
Epoch 194/500
72/72 - 0s - loss: 0.7272 - accuracy: 0.7807 - val_loss: 0.4288 - val_accuracy: 0.8790 - 335ms/epoch - 5ms/step
Epoch 195/500
72/72 - 0s - loss: 0.7258 - accuracy: 0.7808 - val_loss: 0.4224 - val_accuracy: 0.8823 - 334ms/epoch - 5ms/step
Epoch 196/500
72/72 - 0s - loss: 0.7083 - accuracy: 0.7858 - val_loss: 0.4347 - val_accuracy: 0.8777 - 338ms/epoch - 5ms/step
Epoch 197/500
72/72 - 0s - loss: 0.7175 - accuracy: 0.7818 - val_loss: 0.4445 - val_accuracy: 0.8760 - 331ms/epoch - 5ms/step
Epoch 198/500
72/72 - 0s - loss: 0.7142 - accuracy: 0.7807 - val_loss: 0.4546 - val_accuracy: 0.8717 - 339ms/epoch - 5ms/step
Epoch 199/500
72/72 - 0s - loss: 0.7232 - accuracy: 0.7802 - val_loss: 0.4123 - val_accuracy: 0.8817 - 329ms/epoch - 5ms/step
Epoch 200/500
72/72 - 0s - loss: 0.7129 - accuracy: 0.7818 - val_loss: 0.4366 - val_accuracy: 0.8750 - 335ms/epoch - 5ms/step
Epoch 201/500
72/72 - 0s - loss: 0.7088 - accuracy: 0.7832 - val_loss: 0.4181 - val_accuracy: 0.8807 - 334ms/epoch - 5ms/step
Epoch 202/500
72/72 - 0s - loss: 0.7390 - accuracy: 0.7757 - val_loss: 0.4475 - val_accuracy: 0.8680 - 332ms/epoch - 5ms/step
Epoch 203/500
72/72 - 0s - loss: 0.7129 - accuracy: 0.7808 - val_loss: 0.4295 - val_accuracy: 0.8777 - 334ms/epoch - 5ms/step
Epoch 204/500
72/72 - 0s - loss: 0.7144 - accuracy: 0.7766 - val_loss: 0.4364 - val_accuracy: 0.8750 - 344ms/epoch - 5ms/step
Epoch 205/500
72/72 - 0s - loss: 0.7166 - accuracy: 0.7836 - val_loss: 0.4132 - val_accuracy: 0.8847 - 348ms/epoch - 5ms/step
Epoch 206/500
72/72 - 0s - loss: 0.6978 - accuracy: 0.7900 - val_loss: 0.4341 - val_accuracy: 0.8760 - 345ms/epoch - 5ms/step
Epoch 207/500
72/72 - 0s - loss: 0.7064 - accuracy: 0.7854 - val_loss: 0.4342 - val_accuracy: 0.8723 - 340ms/epoch - 5ms/step
Epoch 208/500
72/72 - 0s - loss: 0.7029 - accuracy: 0.7834 - val_loss: 0.4578 - val_accuracy: 0.8693 - 330ms/epoch - 5ms/step
Epoch 209/500
72/72 - 0s - loss: 0.7092 - accuracy: 0.7827 - val_loss: 0.4207 - val_accuracy: 0.8843 - 331ms/epoch - 5ms/step
Epoch 210/500
72/72 - 0s - loss: 0.6812 - accuracy: 0.7897 - val_loss: 0.4425 - val_accuracy: 0.8753 - 336ms/epoch - 5ms/step
Epoch 211/500
72/72 - 0s - loss: 0.6924 - accuracy: 0.7915 - val_loss: 0.4516 - val_accuracy: 0.8683 - 340ms/epoch - 5ms/step
Epoch 212/500
72/72 - 0s - loss: 0.7017 - accuracy: 0.7859 - val_loss: 0.4337 - val_accuracy: 0.8753 - 331ms/epoch - 5ms/step
Epoch 213/500
72/72 - 0s - loss: 0.6975 - accuracy: 0.7909 - val_loss: 0.4397 - val_accuracy: 0.8743 - 334ms/epoch - 5ms/step
Epoch 214/500
72/72 - 0s - loss: 0.6913 - accuracy: 0.7876 - val_loss: 0.4135 - val_accuracy: 0.8853 - 335ms/epoch - 5ms/step
Epoch 215/500
72/72 - 0s - loss: 0.6781 - accuracy: 0.7929 - val_loss: 0.4317 - val_accuracy: 0.8763 - 328ms/epoch - 5ms/step
Epoch 216/500
72/72 - 0s - loss: 0.6689 - accuracy: 0.7962 - val_loss: 0.3954 - val_accuracy: 0.8913 - 338ms/epoch - 5ms/step
Epoch 217/500
72/72 - 0s - loss: 0.6972 - accuracy: 0.7876 - val_loss: 0.4153 - val_accuracy: 0.8850 - 331ms/epoch - 5ms/step
Epoch 218/500
72/72 - 0s - loss: 0.6892 - accuracy: 0.7904 - val_loss: 0.4775 - val_accuracy: 0.8617 - 338ms/epoch - 5ms/step
Epoch 219/500
72/72 - 0s - loss: 0.6814 - accuracy: 0.7957 - val_loss: 0.4073 - val_accuracy: 0.8870 - 348ms/epoch - 5ms/step
Epoch 220/500
72/72 - 0s - loss: 0.6853 - accuracy: 0.7893 - val_loss: 0.4204 - val_accuracy: 0.8810 - 350ms/epoch - 5ms/step
Epoch 221/500
72/72 - 0s - loss: 0.6789 - accuracy: 0.7964 - val_loss: 0.4051 - val_accuracy: 0.8893 - 333ms/epoch - 5ms/step
Epoch 222/500
72/72 - 0s - loss: 0.6888 - accuracy: 0.7879 - val_loss: 0.4002 - val_accuracy: 0.8897 - 333ms/epoch - 5ms/step
Epoch 223/500
72/72 - 0s - loss: 0.6864 - accuracy: 0.7917 - val_loss: 0.4091 - val_accuracy: 0.8870 - 342ms/epoch - 5ms/step
Epoch 224/500
72/72 - 0s - loss: 0.6768 - accuracy: 0.7946 - val_loss: 0.4228 - val_accuracy: 0.8843 - 338ms/epoch - 5ms/step
Epoch 225/500
72/72 - 0s - loss: 0.6778 - accuracy: 0.7943 - val_loss: 0.4291 - val_accuracy: 0.8793 - 344ms/epoch - 5ms/step
Epoch 226/500
72/72 - 0s - loss: 0.6950 - accuracy: 0.7889 - val_loss: 0.3989 - val_accuracy: 0.8877 - 333ms/epoch - 5ms/step
Epoch 227/500
72/72 - 0s - loss: 0.6693 - accuracy: 0.7958 - val_loss: 0.4324 - val_accuracy: 0.8857 - 333ms/epoch - 5ms/step
Epoch 228/500
72/72 - 0s - loss: 0.6660 - accuracy: 0.7995 - val_loss: 0.4045 - val_accuracy: 0.8920 - 339ms/epoch - 5ms/step
Epoch 229/500
72/72 - 0s - loss: 0.6728 - accuracy: 0.7952 - val_loss: 0.4227 - val_accuracy: 0.8817 - 336ms/epoch - 5ms/step
Epoch 230/500
72/72 - 0s - loss: 0.6700 - accuracy: 0.7973 - val_loss: 0.4015 - val_accuracy: 0.8890 - 337ms/epoch - 5ms/step
Epoch 231/500
72/72 - 0s - loss: 0.6584 - accuracy: 0.7988 - val_loss: 0.4725 - val_accuracy: 0.8647 - 334ms/epoch - 5ms/step
Epoch 232/500
72/72 - 0s - loss: 0.6606 - accuracy: 0.7983 - val_loss: 0.3925 - val_accuracy: 0.8930 - 339ms/epoch - 5ms/step
Epoch 233/500
72/72 - 0s - loss: 0.6538 - accuracy: 0.8003 - val_loss: 0.4072 - val_accuracy: 0.8873 - 332ms/epoch - 5ms/step
Epoch 234/500
72/72 - 0s - loss: 0.6524 - accuracy: 0.8007 - val_loss: 0.3954 - val_accuracy: 0.8900 - 338ms/epoch - 5ms/step
Epoch 235/500
72/72 - 0s - loss: 0.6669 - accuracy: 0.7971 - val_loss: 0.4252 - val_accuracy: 0.8833 - 346ms/epoch - 5ms/step
Epoch 236/500
72/72 - 0s - loss: 0.6493 - accuracy: 0.8002 - val_loss: 0.3902 - val_accuracy: 0.8940 - 334ms/epoch - 5ms/step
Epoch 237/500
72/72 - 0s - loss: 0.6495 - accuracy: 0.8022 - val_loss: 0.3852 - val_accuracy: 0.8933 - 332ms/epoch - 5ms/step
Epoch 238/500
72/72 - 0s - loss: 0.6489 - accuracy: 0.8041 - val_loss: 0.4008 - val_accuracy: 0.8873 - 334ms/epoch - 5ms/step
Epoch 239/500
72/72 - 0s - loss: 0.6613 - accuracy: 0.7994 - val_loss: 0.4012 - val_accuracy: 0.8903 - 340ms/epoch - 5ms/step
Epoch 240/500
72/72 - 0s - loss: 0.6545 - accuracy: 0.7990 - val_loss: 0.3955 - val_accuracy: 0.8897 - 335ms/epoch - 5ms/step
Epoch 241/500
72/72 - 0s - loss: 0.6496 - accuracy: 0.8026 - val_loss: 0.4108 - val_accuracy: 0.8853 - 330ms/epoch - 5ms/step
Epoch 242/500
72/72 - 0s - loss: 0.6618 - accuracy: 0.7982 - val_loss: 0.4781 - val_accuracy: 0.8660 - 333ms/epoch - 5ms/step
Epoch 243/500
72/72 - 0s - loss: 0.6491 - accuracy: 0.8033 - val_loss: 0.4714 - val_accuracy: 0.8693 - 332ms/epoch - 5ms/step
Epoch 244/500
72/72 - 0s - loss: 0.6519 - accuracy: 0.8019 - val_loss: 0.4491 - val_accuracy: 0.8700 - 329ms/epoch - 5ms/step
Epoch 245/500
72/72 - 0s - loss: 0.6471 - accuracy: 0.8036 - val_loss: 0.4049 - val_accuracy: 0.8850 - 338ms/epoch - 5ms/step
Epoch 246/500
72/72 - 0s - loss: 0.6534 - accuracy: 0.7991 - val_loss: 0.4061 - val_accuracy: 0.8863 - 346ms/epoch - 5ms/step
Epoch 247/500
72/72 - 0s - loss: 0.6498 - accuracy: 0.8012 - val_loss: 0.3845 - val_accuracy: 0.8953 - 339ms/epoch - 5ms/step
Epoch 248/500
72/72 - 0s - loss: 0.6463 - accuracy: 0.8043 - val_loss: 0.3885 - val_accuracy: 0.8927 - 334ms/epoch - 5ms/step
Epoch 249/500
72/72 - 0s - loss: 0.6302 - accuracy: 0.8067 - val_loss: 0.3882 - val_accuracy: 0.8900 - 338ms/epoch - 5ms/step
Epoch 250/500
72/72 - 0s - loss: 0.6633 - accuracy: 0.8016 - val_loss: 0.4032 - val_accuracy: 0.8917 - 331ms/epoch - 5ms/step
Epoch 251/500
72/72 - 0s - loss: 0.6400 - accuracy: 0.8062 - val_loss: 0.4279 - val_accuracy: 0.8857 - 336ms/epoch - 5ms/step
Epoch 252/500
72/72 - 0s - loss: 0.6434 - accuracy: 0.8054 - val_loss: 0.3861 - val_accuracy: 0.8947 - 337ms/epoch - 5ms/step
Epoch 253/500
72/72 - 0s - loss: 0.6367 - accuracy: 0.8068 - val_loss: 0.3884 - val_accuracy: 0.8950 - 335ms/epoch - 5ms/step
Epoch 254/500
72/72 - 0s - loss: 0.6478 - accuracy: 0.8052 - val_loss: 0.4331 - val_accuracy: 0.8750 - 338ms/epoch - 5ms/step
Epoch 255/500
72/72 - 0s - loss: 0.6314 - accuracy: 0.8087 - val_loss: 0.4187 - val_accuracy: 0.8813 - 335ms/epoch - 5ms/step
Epoch 256/500
72/72 - 0s - loss: 0.6420 - accuracy: 0.8033 - val_loss: 0.3778 - val_accuracy: 0.8960 - 330ms/epoch - 5ms/step
Epoch 257/500
72/72 - 0s - loss: 0.6462 - accuracy: 0.8066 - val_loss: 0.4100 - val_accuracy: 0.8873 - 330ms/epoch - 5ms/step
Epoch 258/500
72/72 - 0s - loss: 0.6350 - accuracy: 0.8038 - val_loss: 0.4180 - val_accuracy: 0.8867 - 336ms/epoch - 5ms/step
Epoch 259/500
72/72 - 0s - loss: 0.6324 - accuracy: 0.8067 - val_loss: 0.4232 - val_accuracy: 0.8763 - 329ms/epoch - 5ms/step
Epoch 260/500
72/72 - 0s - loss: 0.6276 - accuracy: 0.8080 - val_loss: 0.3898 - val_accuracy: 0.8890 - 340ms/epoch - 5ms/step
Epoch 261/500
72/72 - 0s - loss: 0.6385 - accuracy: 0.8077 - val_loss: 0.3711 - val_accuracy: 0.8987 - 335ms/epoch - 5ms/step
Epoch 262/500
72/72 - 0s - loss: 0.6255 - accuracy: 0.8075 - val_loss: 0.4137 - val_accuracy: 0.8843 - 326ms/epoch - 5ms/step
Epoch 263/500
72/72 - 0s - loss: 0.6363 - accuracy: 0.8033 - val_loss: 0.3841 - val_accuracy: 0.8977 - 334ms/epoch - 5ms/step
Epoch 264/500
72/72 - 0s - loss: 0.6345 - accuracy: 0.8094 - val_loss: 0.4131 - val_accuracy: 0.8847 - 332ms/epoch - 5ms/step
Epoch 265/500
72/72 - 0s - loss: 0.6032 - accuracy: 0.8186 - val_loss: 0.4141 - val_accuracy: 0.8857 - 328ms/epoch - 5ms/step
Epoch 266/500
72/72 - 0s - loss: 0.6469 - accuracy: 0.8022 - val_loss: 0.4006 - val_accuracy: 0.8930 - 332ms/epoch - 5ms/step
Epoch 267/500
72/72 - 0s - loss: 0.6235 - accuracy: 0.8065 - val_loss: 0.3859 - val_accuracy: 0.8963 - 326ms/epoch - 5ms/step
Epoch 268/500
72/72 - 0s - loss: 0.6253 - accuracy: 0.8077 - val_loss: 0.3833 - val_accuracy: 0.8960 - 329ms/epoch - 5ms/step
Epoch 269/500
72/72 - 0s - loss: 0.6271 - accuracy: 0.8097 - val_loss: 0.3697 - val_accuracy: 0.8950 - 333ms/epoch - 5ms/step
Epoch 270/500
72/72 - 0s - loss: 0.6275 - accuracy: 0.8091 - val_loss: 0.3821 - val_accuracy: 0.8947 - 330ms/epoch - 5ms/step
Epoch 271/500
72/72 - 0s - loss: 0.6385 - accuracy: 0.8071 - val_loss: 0.3830 - val_accuracy: 0.8970 - 329ms/epoch - 5ms/step
Epoch 272/500
72/72 - 0s - loss: 0.6109 - accuracy: 0.8153 - val_loss: 0.3901 - val_accuracy: 0.8903 - 329ms/epoch - 5ms/step
Epoch 273/500
72/72 - 0s - loss: 0.6256 - accuracy: 0.8119 - val_loss: 0.3839 - val_accuracy: 0.8910 - 330ms/epoch - 5ms/step
Epoch 274/500
72/72 - 0s - loss: 0.6166 - accuracy: 0.8144 - val_loss: 0.3706 - val_accuracy: 0.8937 - 327ms/epoch - 5ms/step
Epoch 275/500
72/72 - 0s - loss: 0.6226 - accuracy: 0.8132 - val_loss: 0.3920 - val_accuracy: 0.8920 - 335ms/epoch - 5ms/step
Epoch 276/500
72/72 - 0s - loss: 0.6146 - accuracy: 0.8108 - val_loss: 0.3692 - val_accuracy: 0.9020 - 328ms/epoch - 5ms/step
Epoch 277/500
72/72 - 0s - loss: 0.6178 - accuracy: 0.8089 - val_loss: 0.3815 - val_accuracy: 0.8970 - 326ms/epoch - 5ms/step
Epoch 278/500
72/72 - 0s - loss: 0.6074 - accuracy: 0.8158 - val_loss: 0.3688 - val_accuracy: 0.8960 - 330ms/epoch - 5ms/step
Epoch 279/500
72/72 - 0s - loss: 0.6242 - accuracy: 0.8099 - val_loss: 0.3752 - val_accuracy: 0.8940 - 329ms/epoch - 5ms/step
Epoch 280/500
72/72 - 0s - loss: 0.6296 - accuracy: 0.8091 - val_loss: 0.3619 - val_accuracy: 0.8977 - 328ms/epoch - 5ms/step
Epoch 281/500
72/72 - 0s - loss: 0.6219 - accuracy: 0.8103 - val_loss: 0.3879 - val_accuracy: 0.8917 - 328ms/epoch - 5ms/step
Epoch 282/500
72/72 - 0s - loss: 0.6067 - accuracy: 0.8160 - val_loss: 0.3682 - val_accuracy: 0.8977 - 331ms/epoch - 5ms/step
Epoch 283/500
72/72 - 0s - loss: 0.6125 - accuracy: 0.8151 - val_loss: 0.3917 - val_accuracy: 0.8887 - 331ms/epoch - 5ms/step
Epoch 284/500
72/72 - 0s - loss: 0.6218 - accuracy: 0.8164 - val_loss: 0.3856 - val_accuracy: 0.8930 - 333ms/epoch - 5ms/step
Epoch 285/500
72/72 - 0s - loss: 0.6364 - accuracy: 0.8094 - val_loss: 0.4113 - val_accuracy: 0.8853 - 345ms/epoch - 5ms/step
Epoch 286/500
72/72 - 0s - loss: 0.6128 - accuracy: 0.8126 - val_loss: 0.3759 - val_accuracy: 0.8953 - 331ms/epoch - 5ms/step
Epoch 287/500
72/72 - 0s - loss: 0.6114 - accuracy: 0.8163 - val_loss: 0.3887 - val_accuracy: 0.8930 - 339ms/epoch - 5ms/step
Epoch 288/500
72/72 - 0s - loss: 0.6086 - accuracy: 0.8143 - val_loss: 0.3771 - val_accuracy: 0.8940 - 333ms/epoch - 5ms/step
Epoch 289/500
72/72 - 0s - loss: 0.6041 - accuracy: 0.8152 - val_loss: 0.3807 - val_accuracy: 0.8970 - 334ms/epoch - 5ms/step
Epoch 290/500
72/72 - 0s - loss: 0.6049 - accuracy: 0.8167 - val_loss: 0.4332 - val_accuracy: 0.8807 - 336ms/epoch - 5ms/step
Epoch 291/500
72/72 - 0s - loss: 0.6073 - accuracy: 0.8182 - val_loss: 0.3536 - val_accuracy: 0.9050 - 333ms/epoch - 5ms/step
Epoch 292/500
72/72 - 0s - loss: 0.6036 - accuracy: 0.8186 - val_loss: 0.3832 - val_accuracy: 0.8960 - 334ms/epoch - 5ms/step
Epoch 293/500
72/72 - 0s - loss: 0.6091 - accuracy: 0.8132 - val_loss: 0.3579 - val_accuracy: 0.8980 - 346ms/epoch - 5ms/step
Epoch 294/500
72/72 - 0s - loss: 0.5925 - accuracy: 0.8179 - val_loss: 0.3860 - val_accuracy: 0.8943 - 347ms/epoch - 5ms/step
Epoch 295/500
72/72 - 0s - loss: 0.6217 - accuracy: 0.8119 - val_loss: 0.3715 - val_accuracy: 0.8947 - 345ms/epoch - 5ms/step
Epoch 296/500
72/72 - 0s - loss: 0.6062 - accuracy: 0.8177 - val_loss: 0.3909 - val_accuracy: 0.8917 - 336ms/epoch - 5ms/step
Epoch 297/500
72/72 - 0s - loss: 0.5970 - accuracy: 0.8212 - val_loss: 0.3723 - val_accuracy: 0.9003 - 331ms/epoch - 5ms/step
Epoch 298/500
72/72 - 0s - loss: 0.5974 - accuracy: 0.8209 - val_loss: 0.3535 - val_accuracy: 0.9040 - 342ms/epoch - 5ms/step
Epoch 299/500
72/72 - 0s - loss: 0.6105 - accuracy: 0.8124 - val_loss: 0.3523 - val_accuracy: 0.9057 - 339ms/epoch - 5ms/step
Epoch 300/500
72/72 - 0s - loss: 0.6047 - accuracy: 0.8190 - val_loss: 0.3556 - val_accuracy: 0.9007 - 331ms/epoch - 5ms/step
Epoch 301/500
72/72 - 0s - loss: 0.5882 - accuracy: 0.8183 - val_loss: 0.3590 - val_accuracy: 0.9040 - 330ms/epoch - 5ms/step
Epoch 302/500
72/72 - 0s - loss: 0.5922 - accuracy: 0.8218 - val_loss: 0.3661 - val_accuracy: 0.8993 - 336ms/epoch - 5ms/step
Epoch 303/500
72/72 - 0s - loss: 0.6057 - accuracy: 0.8147 - val_loss: 0.3976 - val_accuracy: 0.8900 - 329ms/epoch - 5ms/step
Epoch 304/500
72/72 - 0s - loss: 0.5849 - accuracy: 0.8223 - val_loss: 0.3857 - val_accuracy: 0.8927 - 335ms/epoch - 5ms/step
Epoch 305/500
72/72 - 0s - loss: 0.5890 - accuracy: 0.8219 - val_loss: 0.3718 - val_accuracy: 0.8980 - 334ms/epoch - 5ms/step
Epoch 306/500
72/72 - 0s - loss: 0.5917 - accuracy: 0.8195 - val_loss: 0.3713 - val_accuracy: 0.9017 - 333ms/epoch - 5ms/step
Epoch 307/500
72/72 - 0s - loss: 0.5932 - accuracy: 0.8185 - val_loss: 0.3599 - val_accuracy: 0.8997 - 334ms/epoch - 5ms/step
Epoch 308/500
72/72 - 0s - loss: 0.5891 - accuracy: 0.8200 - val_loss: 0.3549 - val_accuracy: 0.9053 - 332ms/epoch - 5ms/step
Epoch 309/500
72/72 - 0s - loss: 0.5943 - accuracy: 0.8159 - val_loss: 0.3785 - val_accuracy: 0.8930 - 335ms/epoch - 5ms/step
Epoch 310/500
72/72 - 0s - loss: 0.6011 - accuracy: 0.8175 - val_loss: 0.4059 - val_accuracy: 0.8830 - 343ms/epoch - 5ms/step
Epoch 311/500
72/72 - 0s - loss: 0.5832 - accuracy: 0.8239 - val_loss: 0.3577 - val_accuracy: 0.8963 - 351ms/epoch - 5ms/step
Epoch 312/500
72/72 - 0s - loss: 0.5880 - accuracy: 0.8212 - val_loss: 0.4083 - val_accuracy: 0.8843 - 341ms/epoch - 5ms/step
Epoch 313/500
72/72 - 0s - loss: 0.5826 - accuracy: 0.8212 - val_loss: 0.3734 - val_accuracy: 0.8947 - 340ms/epoch - 5ms/step
Epoch 314/500
72/72 - 0s - loss: 0.6045 - accuracy: 0.8146 - val_loss: 0.4916 - val_accuracy: 0.8587 - 336ms/epoch - 5ms/step
Epoch 315/500
72/72 - 0s - loss: 0.5830 - accuracy: 0.8242 - val_loss: 0.3622 - val_accuracy: 0.8977 - 330ms/epoch - 5ms/step
Epoch 316/500
72/72 - 0s - loss: 0.5902 - accuracy: 0.8232 - val_loss: 0.3595 - val_accuracy: 0.8993 - 335ms/epoch - 5ms/step
Epoch 317/500
72/72 - 0s - loss: 0.5789 - accuracy: 0.8218 - val_loss: 0.3754 - val_accuracy: 0.8917 - 332ms/epoch - 5ms/step
Epoch 318/500
72/72 - 0s - loss: 0.5949 - accuracy: 0.8192 - val_loss: 0.3717 - val_accuracy: 0.8973 - 333ms/epoch - 5ms/step
Epoch 319/500
72/72 - 0s - loss: 0.5654 - accuracy: 0.8272 - val_loss: 0.3529 - val_accuracy: 0.9023 - 341ms/epoch - 5ms/step
Epoch 320/500
72/72 - 0s - loss: 0.5703 - accuracy: 0.8312 - val_loss: 0.3741 - val_accuracy: 0.8933 - 334ms/epoch - 5ms/step
Epoch 321/500
72/72 - 0s - loss: 0.5863 - accuracy: 0.8209 - val_loss: 0.3988 - val_accuracy: 0.8883 - 331ms/epoch - 5ms/step
Epoch 322/500
72/72 - 0s - loss: 0.5821 - accuracy: 0.8258 - val_loss: 0.3602 - val_accuracy: 0.9003 - 332ms/epoch - 5ms/step
Epoch 323/500
72/72 - 0s - loss: 0.5871 - accuracy: 0.8212 - val_loss: 0.3540 - val_accuracy: 0.9020 - 331ms/epoch - 5ms/step
Epoch 324/500
72/72 - 0s - loss: 0.5847 - accuracy: 0.8215 - val_loss: 0.3520 - val_accuracy: 0.9050 - 334ms/epoch - 5ms/step
Epoch 325/500
72/72 - 0s - loss: 0.5711 - accuracy: 0.8269 - val_loss: 0.3423 - val_accuracy: 0.9070 - 348ms/epoch - 5ms/step
Epoch 326/500
72/72 - 0s - loss: 0.5794 - accuracy: 0.8223 - val_loss: 0.3462 - val_accuracy: 0.9040 - 354ms/epoch - 5ms/step
Epoch 327/500
72/72 - 0s - loss: 0.5777 - accuracy: 0.8250 - val_loss: 0.3609 - val_accuracy: 0.9000 - 359ms/epoch - 5ms/step
Epoch 328/500
72/72 - 0s - loss: 0.5740 - accuracy: 0.8305 - val_loss: 0.3776 - val_accuracy: 0.8947 - 352ms/epoch - 5ms/step
Epoch 329/500
72/72 - 0s - loss: 0.5676 - accuracy: 0.8293 - val_loss: 0.3715 - val_accuracy: 0.8957 - 348ms/epoch - 5ms/step
Epoch 330/500
72/72 - 0s - loss: 0.5643 - accuracy: 0.8299 - val_loss: 0.3570 - val_accuracy: 0.8993 - 347ms/epoch - 5ms/step
Epoch 331/500
72/72 - 0s - loss: 0.5717 - accuracy: 0.8268 - val_loss: 0.3863 - val_accuracy: 0.8963 - 355ms/epoch - 5ms/step
Epoch 332/500
72/72 - 0s - loss: 0.5897 - accuracy: 0.8237 - val_loss: 0.3694 - val_accuracy: 0.8980 - 343ms/epoch - 5ms/step
Epoch 333/500
72/72 - 0s - loss: 0.5787 - accuracy: 0.8255 - val_loss: 0.3417 - val_accuracy: 0.9100 - 342ms/epoch - 5ms/step
Epoch 334/500
72/72 - 0s - loss: 0.5751 - accuracy: 0.8273 - val_loss: 0.3491 - val_accuracy: 0.9037 - 348ms/epoch - 5ms/step
Epoch 335/500
72/72 - 0s - loss: 0.5894 - accuracy: 0.8217 - val_loss: 0.3765 - val_accuracy: 0.8970 - 351ms/epoch - 5ms/step
Epoch 336/500
72/72 - 0s - loss: 0.5783 - accuracy: 0.8251 - val_loss: 0.3992 - val_accuracy: 0.8920 - 349ms/epoch - 5ms/step
Epoch 337/500
72/72 - 0s - loss: 0.5724 - accuracy: 0.8274 - val_loss: 0.4380 - val_accuracy: 0.8737 - 339ms/epoch - 5ms/step
Epoch 338/500
72/72 - 0s - loss: 0.5635 - accuracy: 0.8267 - val_loss: 0.3615 - val_accuracy: 0.9030 - 336ms/epoch - 5ms/step
Epoch 339/500
72/72 - 0s - loss: 0.5752 - accuracy: 0.8257 - val_loss: 0.3931 - val_accuracy: 0.8897 - 337ms/epoch - 5ms/step
Epoch 340/500
72/72 - 0s - loss: 0.5758 - accuracy: 0.8275 - val_loss: 0.3544 - val_accuracy: 0.8993 - 337ms/epoch - 5ms/step
Epoch 341/500
72/72 - 0s - loss: 0.5610 - accuracy: 0.8273 - val_loss: 0.3652 - val_accuracy: 0.8977 - 335ms/epoch - 5ms/step
Epoch 342/500
72/72 - 0s - loss: 0.5621 - accuracy: 0.8284 - val_loss: 0.3487 - val_accuracy: 0.9023 - 337ms/epoch - 5ms/step
Epoch 343/500
72/72 - 0s - loss: 0.5591 - accuracy: 0.8314 - val_loss: 0.3518 - val_accuracy: 0.9000 - 334ms/epoch - 5ms/step
Epoch 344/500
72/72 - 0s - loss: 0.5631 - accuracy: 0.8314 - val_loss: 0.3589 - val_accuracy: 0.9010 - 341ms/epoch - 5ms/step
Epoch 345/500
72/72 - 0s - loss: 0.5486 - accuracy: 0.8323 - val_loss: 0.4059 - val_accuracy: 0.8847 - 333ms/epoch - 5ms/step
Epoch 346/500
72/72 - 0s - loss: 0.5562 - accuracy: 0.8327 - val_loss: 0.3570 - val_accuracy: 0.8993 - 336ms/epoch - 5ms/step
Epoch 347/500
72/72 - 0s - loss: 0.5551 - accuracy: 0.8323 - val_loss: 0.3466 - val_accuracy: 0.9023 - 340ms/epoch - 5ms/step
Epoch 348/500
72/72 - 0s - loss: 0.5624 - accuracy: 0.8324 - val_loss: 0.3613 - val_accuracy: 0.8983 - 338ms/epoch - 5ms/step
Epoch 349/500
72/72 - 0s - loss: 0.5660 - accuracy: 0.8309 - val_loss: 0.3703 - val_accuracy: 0.8960 - 335ms/epoch - 5ms/step
Epoch 350/500
72/72 - 0s - loss: 0.5583 - accuracy: 0.8279 - val_loss: 0.3557 - val_accuracy: 0.8957 - 335ms/epoch - 5ms/step
Epoch 351/500
72/72 - 0s - loss: 0.5595 - accuracy: 0.8290 - val_loss: 0.3618 - val_accuracy: 0.8967 - 337ms/epoch - 5ms/step
Epoch 352/500
72/72 - 0s - loss: 0.5579 - accuracy: 0.8297 - val_loss: 0.3622 - val_accuracy: 0.9000 - 332ms/epoch - 5ms/step
Epoch 353/500
72/72 - 0s - loss: 0.5492 - accuracy: 0.8357 - val_loss: 0.3854 - val_accuracy: 0.8940 - 332ms/epoch - 5ms/step
Epoch 354/500
72/72 - 0s - loss: 0.5530 - accuracy: 0.8312 - val_loss: 0.3485 - val_accuracy: 0.9057 - 335ms/epoch - 5ms/step
Epoch 355/500
72/72 - 0s - loss: 0.5505 - accuracy: 0.8297 - val_loss: 0.3818 - val_accuracy: 0.8963 - 328ms/epoch - 5ms/step
Epoch 356/500
72/72 - 0s - loss: 0.5457 - accuracy: 0.8362 - val_loss: 0.4048 - val_accuracy: 0.8823 - 333ms/epoch - 5ms/step
Epoch 357/500
72/72 - 0s - loss: 0.5845 - accuracy: 0.8241 - val_loss: 0.3644 - val_accuracy: 0.8973 - 341ms/epoch - 5ms/step
Epoch 358/500
72/72 - 0s - loss: 0.5550 - accuracy: 0.8307 - val_loss: 0.3589 - val_accuracy: 0.9023 - 329ms/epoch - 5ms/step
Epoch 359/500
72/72 - 0s - loss: 0.5438 - accuracy: 0.8369 - val_loss: 0.3784 - val_accuracy: 0.8963 - 333ms/epoch - 5ms/step
Epoch 360/500
72/72 - 0s - loss: 0.5621 - accuracy: 0.8302 - val_loss: 0.3624 - val_accuracy: 0.8990 - 334ms/epoch - 5ms/step
Epoch 361/500
72/72 - 0s - loss: 0.5664 - accuracy: 0.8302 - val_loss: 0.3833 - val_accuracy: 0.8957 - 331ms/epoch - 5ms/step
Epoch 362/500
72/72 - 0s - loss: 0.5510 - accuracy: 0.8346 - val_loss: 0.3919 - val_accuracy: 0.8870 - 333ms/epoch - 5ms/step
Epoch 363/500
72/72 - 0s - loss: 0.5402 - accuracy: 0.8340 - val_loss: 0.3730 - val_accuracy: 0.8920 - 339ms/epoch - 5ms/step
Epoch 364/500
72/72 - 0s - loss: 0.5573 - accuracy: 0.8281 - val_loss: 0.3618 - val_accuracy: 0.9000 - 333ms/epoch - 5ms/step
Epoch 365/500
72/72 - 0s - loss: 0.5524 - accuracy: 0.8302 - val_loss: 0.3437 - val_accuracy: 0.9047 - 333ms/epoch - 5ms/step
Epoch 366/500
72/72 - 0s - loss: 0.5482 - accuracy: 0.8314 - val_loss: 0.3628 - val_accuracy: 0.9003 - 341ms/epoch - 5ms/step
Epoch 367/500
72/72 - 0s - loss: 0.5405 - accuracy: 0.8370 - val_loss: 0.3479 - val_accuracy: 0.9047 - 334ms/epoch - 5ms/step
Epoch 368/500
72/72 - 0s - loss: 0.5650 - accuracy: 0.8293 - val_loss: 0.3730 - val_accuracy: 0.8970 - 333ms/epoch - 5ms/step
Epoch 369/500
72/72 - 0s - loss: 0.5666 - accuracy: 0.8279 - val_loss: 0.3588 - val_accuracy: 0.9047 - 337ms/epoch - 5ms/step
Epoch 370/500
72/72 - 0s - loss: 0.5441 - accuracy: 0.8336 - val_loss: 0.3820 - val_accuracy: 0.8953 - 330ms/epoch - 5ms/step
Epoch 371/500
72/72 - 0s - loss: 0.5476 - accuracy: 0.8330 - val_loss: 0.3661 - val_accuracy: 0.9007 - 336ms/epoch - 5ms/step
Epoch 372/500
72/72 - 0s - loss: 0.5461 - accuracy: 0.8341 - val_loss: 0.3865 - val_accuracy: 0.8910 - 338ms/epoch - 5ms/step
Epoch 373/500
72/72 - 0s - loss: 0.5483 - accuracy: 0.8343 - val_loss: 0.3516 - val_accuracy: 0.9020 - 332ms/epoch - 5ms/step
Epoch 374/500
72/72 - 0s - loss: 0.5507 - accuracy: 0.8334 - val_loss: 0.3431 - val_accuracy: 0.9023 - 330ms/epoch - 5ms/step
Epoch 375/500
72/72 - 0s - loss: 0.5505 - accuracy: 0.8347 - val_loss: 0.3578 - val_accuracy: 0.9020 - 334ms/epoch - 5ms/step
Epoch 376/500
72/72 - 0s - loss: 0.5474 - accuracy: 0.8355 - val_loss: 0.3643 - val_accuracy: 0.9033 - 338ms/epoch - 5ms/step
Epoch 377/500
72/72 - 0s - loss: 0.5451 - accuracy: 0.8334 - val_loss: 0.3847 - val_accuracy: 0.8923 - 336ms/epoch - 5ms/step
Epoch 378/500
72/72 - 0s - loss: 0.5344 - accuracy: 0.8393 - val_loss: 0.3270 - val_accuracy: 0.9063 - 336ms/epoch - 5ms/step
Epoch 379/500
72/72 - 0s - loss: 0.5440 - accuracy: 0.8351 - val_loss: 0.3471 - val_accuracy: 0.9023 - 331ms/epoch - 5ms/step
Epoch 380/500
72/72 - 0s - loss: 0.5449 - accuracy: 0.8365 - val_loss: 0.3350 - val_accuracy: 0.9067 - 333ms/epoch - 5ms/step
Epoch 381/500
72/72 - 0s - loss: 0.5507 - accuracy: 0.8318 - val_loss: 0.3886 - val_accuracy: 0.8920 - 346ms/epoch - 5ms/step
Epoch 382/500
72/72 - 0s - loss: 0.5338 - accuracy: 0.8413 - val_loss: 0.3512 - val_accuracy: 0.9017 - 341ms/epoch - 5ms/step
Epoch 383/500
72/72 - 0s - loss: 0.5331 - accuracy: 0.8395 - val_loss: 0.3432 - val_accuracy: 0.9050 - 340ms/epoch - 5ms/step
Epoch 384/500
72/72 - 0s - loss: 0.5300 - accuracy: 0.8371 - val_loss: 0.3719 - val_accuracy: 0.8953 - 340ms/epoch - 5ms/step
Epoch 385/500
72/72 - 0s - loss: 0.5351 - accuracy: 0.8355 - val_loss: 0.3595 - val_accuracy: 0.8990 - 335ms/epoch - 5ms/step
Epoch 386/500
72/72 - 0s - loss: 0.5366 - accuracy: 0.8413 - val_loss: 0.3400 - val_accuracy: 0.9063 - 338ms/epoch - 5ms/step
Epoch 387/500
72/72 - 0s - loss: 0.5378 - accuracy: 0.8371 - val_loss: 0.3592 - val_accuracy: 0.8953 - 339ms/epoch - 5ms/step
Epoch 388/500
72/72 - 0s - loss: 0.5516 - accuracy: 0.8325 - val_loss: 0.3988 - val_accuracy: 0.8867 - 335ms/epoch - 5ms/step
Epoch 389/500
72/72 - 0s - loss: 0.5333 - accuracy: 0.8406 - val_loss: 0.3321 - val_accuracy: 0.9047 - 336ms/epoch - 5ms/step
Epoch 390/500
72/72 - 0s - loss: 0.5442 - accuracy: 0.8331 - val_loss: 0.3677 - val_accuracy: 0.8967 - 336ms/epoch - 5ms/step
Epoch 391/500
72/72 - 0s - loss: 0.5546 - accuracy: 0.8311 - val_loss: 0.3646 - val_accuracy: 0.8970 - 333ms/epoch - 5ms/step
Epoch 392/500
72/72 - 0s - loss: 0.5373 - accuracy: 0.8363 - val_loss: 0.3492 - val_accuracy: 0.9000 - 329ms/epoch - 5ms/step
Epoch 393/500
72/72 - 0s - loss: 0.5278 - accuracy: 0.8415 - val_loss: 0.3611 - val_accuracy: 0.8993 - 335ms/epoch - 5ms/step
Epoch 394/500
72/72 - 0s - loss: 0.5265 - accuracy: 0.8407 - val_loss: 0.3490 - val_accuracy: 0.9063 - 335ms/epoch - 5ms/step
Epoch 395/500
72/72 - 0s - loss: 0.5451 - accuracy: 0.8363 - val_loss: 0.3451 - val_accuracy: 0.9063 - 336ms/epoch - 5ms/step
Epoch 396/500
72/72 - 0s - loss: 0.5366 - accuracy: 0.8378 - val_loss: 0.3562 - val_accuracy: 0.9007 - 339ms/epoch - 5ms/step
Epoch 397/500
72/72 - 0s - loss: 0.5273 - accuracy: 0.8408 - val_loss: 0.3593 - val_accuracy: 0.9010 - 336ms/epoch - 5ms/step
Epoch 398/500
72/72 - 0s - loss: 0.5210 - accuracy: 0.8437 - val_loss: 0.3347 - val_accuracy: 0.9043 - 333ms/epoch - 5ms/step
Epoch 399/500
72/72 - 0s - loss: 0.5307 - accuracy: 0.8385 - val_loss: 0.3858 - val_accuracy: 0.8933 - 335ms/epoch - 5ms/step
Epoch 400/500
72/72 - 0s - loss: 0.5456 - accuracy: 0.8343 - val_loss: 0.3531 - val_accuracy: 0.8993 - 345ms/epoch - 5ms/step
Epoch 401/500
72/72 - 0s - loss: 0.5457 - accuracy: 0.8377 - val_loss: 0.3421 - val_accuracy: 0.9047 - 355ms/epoch - 5ms/step
Epoch 402/500
72/72 - 0s - loss: 0.5309 - accuracy: 0.8367 - val_loss: 0.3503 - val_accuracy: 0.9023 - 352ms/epoch - 5ms/step
Epoch 403/500
72/72 - 0s - loss: 0.5335 - accuracy: 0.8385 - val_loss: 0.3741 - val_accuracy: 0.8917 - 336ms/epoch - 5ms/step
Epoch 404/500
72/72 - 0s - loss: 0.5276 - accuracy: 0.8405 - val_loss: 0.3366 - val_accuracy: 0.9027 - 337ms/epoch - 5ms/step
Epoch 405/500
72/72 - 0s - loss: 0.5402 - accuracy: 0.8368 - val_loss: 0.3708 - val_accuracy: 0.8933 - 337ms/epoch - 5ms/step
Epoch 406/500
72/72 - 0s - loss: 0.5345 - accuracy: 0.8399 - val_loss: 0.3565 - val_accuracy: 0.9010 - 335ms/epoch - 5ms/step
Epoch 407/500
72/72 - 0s - loss: 0.5286 - accuracy: 0.8413 - val_loss: 0.3379 - val_accuracy: 0.9123 - 342ms/epoch - 5ms/step
Epoch 408/500
72/72 - 0s - loss: 0.5341 - accuracy: 0.8394 - val_loss: 0.3637 - val_accuracy: 0.8947 - 333ms/epoch - 5ms/step
Epoch 409/500
72/72 - 0s - loss: 0.5251 - accuracy: 0.8415 - val_loss: 0.3376 - val_accuracy: 0.9053 - 337ms/epoch - 5ms/step
Epoch 410/500
72/72 - 0s - loss: 0.5362 - accuracy: 0.8417 - val_loss: 0.3910 - val_accuracy: 0.8883 - 337ms/epoch - 5ms/step
Epoch 411/500
72/72 - 0s - loss: 0.5324 - accuracy: 0.8361 - val_loss: 0.3565 - val_accuracy: 0.9027 - 331ms/epoch - 5ms/step
Epoch 412/500
72/72 - 0s - loss: 0.5272 - accuracy: 0.8417 - val_loss: 0.3376 - val_accuracy: 0.8990 - 346ms/epoch - 5ms/step
Epoch 413/500
72/72 - 0s - loss: 0.5340 - accuracy: 0.8383 - val_loss: 0.3610 - val_accuracy: 0.8993 - 337ms/epoch - 5ms/step
Epoch 414/500
72/72 - 0s - loss: 0.5230 - accuracy: 0.8416 - val_loss: 0.3491 - val_accuracy: 0.9010 - 331ms/epoch - 5ms/step
Epoch 415/500
72/72 - 0s - loss: 0.5436 - accuracy: 0.8401 - val_loss: 0.3718 - val_accuracy: 0.8953 - 337ms/epoch - 5ms/step
Epoch 416/500
72/72 - 0s - loss: 0.5220 - accuracy: 0.8429 - val_loss: 0.3379 - val_accuracy: 0.9047 - 340ms/epoch - 5ms/step
Epoch 417/500
72/72 - 0s - loss: 0.5339 - accuracy: 0.8394 - val_loss: 0.3262 - val_accuracy: 0.9083 - 333ms/epoch - 5ms/step
Epoch 418/500
72/72 - 0s - loss: 0.5341 - accuracy: 0.8371 - val_loss: 0.3549 - val_accuracy: 0.8977 - 334ms/epoch - 5ms/step
Epoch 419/500
72/72 - 0s - loss: 0.5359 - accuracy: 0.8407 - val_loss: 0.3700 - val_accuracy: 0.8947 - 334ms/epoch - 5ms/step
Epoch 420/500
72/72 - 0s - loss: 0.5354 - accuracy: 0.8385 - val_loss: 0.3778 - val_accuracy: 0.8917 - 333ms/epoch - 5ms/step
Epoch 421/500
72/72 - 0s - loss: 0.5263 - accuracy: 0.8413 - val_loss: 0.3770 - val_accuracy: 0.8960 - 332ms/epoch - 5ms/step
Epoch 422/500
72/72 - 0s - loss: 0.5385 - accuracy: 0.8378 - val_loss: 0.3474 - val_accuracy: 0.9043 - 338ms/epoch - 5ms/step
Epoch 423/500
72/72 - 0s - loss: 0.5173 - accuracy: 0.8430 - val_loss: 0.3372 - val_accuracy: 0.9073 - 331ms/epoch - 5ms/step
Epoch 424/500
72/72 - 0s - loss: 0.5262 - accuracy: 0.8433 - val_loss: 0.3446 - val_accuracy: 0.9037 - 335ms/epoch - 5ms/step
Epoch 425/500
72/72 - 0s - loss: 0.5224 - accuracy: 0.8425 - val_loss: 0.3546 - val_accuracy: 0.8977 - 335ms/epoch - 5ms/step
Epoch 426/500
72/72 - 0s - loss: 0.5241 - accuracy: 0.8414 - val_loss: 0.3748 - val_accuracy: 0.8893 - 344ms/epoch - 5ms/step
Epoch 427/500
72/72 - 0s - loss: 0.5078 - accuracy: 0.8440 - val_loss: 0.3818 - val_accuracy: 0.8933 - 337ms/epoch - 5ms/step
Epoch 428/500
72/72 - 0s - loss: 0.5253 - accuracy: 0.8401 - val_loss: 0.3831 - val_accuracy: 0.8897 - 333ms/epoch - 5ms/step
Epoch 429/500
72/72 - 0s - loss: 0.5350 - accuracy: 0.8378 - val_loss: 0.3421 - val_accuracy: 0.9057 - 329ms/epoch - 5ms/step
Epoch 430/500
72/72 - 0s - loss: 0.5178 - accuracy: 0.8430 - val_loss: 0.3395 - val_accuracy: 0.9070 - 341ms/epoch - 5ms/step
Epoch 431/500
72/72 - 0s - loss: 0.5148 - accuracy: 0.8429 - val_loss: 0.4147 - val_accuracy: 0.8860 - 335ms/epoch - 5ms/step
Epoch 432/500
72/72 - 0s - loss: 0.5162 - accuracy: 0.8413 - val_loss: 0.3460 - val_accuracy: 0.8997 - 335ms/epoch - 5ms/step
Epoch 433/500
72/72 - 0s - loss: 0.5234 - accuracy: 0.8463 - val_loss: 0.3741 - val_accuracy: 0.8940 - 329ms/epoch - 5ms/step
Epoch 434/500
72/72 - 0s - loss: 0.5266 - accuracy: 0.8398 - val_loss: 0.3547 - val_accuracy: 0.9010 - 332ms/epoch - 5ms/step
Epoch 435/500
72/72 - 0s - loss: 0.5140 - accuracy: 0.8475 - val_loss: 0.3305 - val_accuracy: 0.9070 - 329ms/epoch - 5ms/step
Epoch 436/500
72/72 - 0s - loss: 0.5222 - accuracy: 0.8397 - val_loss: 0.3691 - val_accuracy: 0.8950 - 334ms/epoch - 5ms/step
Epoch 437/500
72/72 - 0s - loss: 0.5006 - accuracy: 0.8473 - val_loss: 0.3323 - val_accuracy: 0.9053 - 336ms/epoch - 5ms/step
Epoch 438/500
72/72 - 0s - loss: 0.5241 - accuracy: 0.8389 - val_loss: 0.3652 - val_accuracy: 0.8987 - 335ms/epoch - 5ms/step
Epoch 439/500
72/72 - 0s - loss: 0.5041 - accuracy: 0.8451 - val_loss: 0.3768 - val_accuracy: 0.8897 - 332ms/epoch - 5ms/step
Epoch 440/500
72/72 - 0s - loss: 0.5109 - accuracy: 0.8459 - val_loss: 0.3755 - val_accuracy: 0.8967 - 335ms/epoch - 5ms/step
Epoch 441/500
72/72 - 0s - loss: 0.5167 - accuracy: 0.8434 - val_loss: 0.3397 - val_accuracy: 0.9017 - 331ms/epoch - 5ms/step
Epoch 442/500
72/72 - 0s - loss: 0.5180 - accuracy: 0.8454 - val_loss: 0.3473 - val_accuracy: 0.9053 - 332ms/epoch - 5ms/step
Epoch 443/500
72/72 - 0s - loss: 0.5081 - accuracy: 0.8480 - val_loss: 0.3648 - val_accuracy: 0.8973 - 333ms/epoch - 5ms/step
Epoch 444/500
72/72 - 0s - loss: 0.5066 - accuracy: 0.8414 - val_loss: 0.3972 - val_accuracy: 0.8953 - 332ms/epoch - 5ms/step
Epoch 445/500
72/72 - 0s - loss: 0.5103 - accuracy: 0.8454 - val_loss: 0.3418 - val_accuracy: 0.9070 - 334ms/epoch - 5ms/step
Epoch 446/500
72/72 - 0s - loss: 0.5185 - accuracy: 0.8454 - val_loss: 0.3535 - val_accuracy: 0.8997 - 339ms/epoch - 5ms/step
Epoch 447/500
72/72 - 0s - loss: 0.5050 - accuracy: 0.8489 - val_loss: 0.3422 - val_accuracy: 0.9020 - 331ms/epoch - 5ms/step
Epoch 448/500
72/72 - 0s - loss: 0.5129 - accuracy: 0.8470 - val_loss: 0.3421 - val_accuracy: 0.9067 - 333ms/epoch - 5ms/step
Epoch 449/500
72/72 - 0s - loss: 0.5049 - accuracy: 0.8505 - val_loss: 0.3264 - val_accuracy: 0.9083 - 331ms/epoch - 5ms/step
Epoch 450/500
72/72 - 0s - loss: 0.4957 - accuracy: 0.8493 - val_loss: 0.3411 - val_accuracy: 0.9057 - 332ms/epoch - 5ms/step
Epoch 451/500
72/72 - 0s - loss: 0.5005 - accuracy: 0.8461 - val_loss: 0.3622 - val_accuracy: 0.8983 - 331ms/epoch - 5ms/step
Epoch 452/500
72/72 - 0s - loss: 0.5030 - accuracy: 0.8461 - val_loss: 0.3333 - val_accuracy: 0.9057 - 336ms/epoch - 5ms/step
Epoch 453/500
72/72 - 0s - loss: 0.5070 - accuracy: 0.8479 - val_loss: 0.3503 - val_accuracy: 0.8997 - 337ms/epoch - 5ms/step
Epoch 454/500
72/72 - 0s - loss: 0.5115 - accuracy: 0.8477 - val_loss: 0.3915 - val_accuracy: 0.8953 - 333ms/epoch - 5ms/step
Epoch 455/500
72/72 - 0s - loss: 0.5087 - accuracy: 0.8475 - val_loss: 0.3825 - val_accuracy: 0.8910 - 334ms/epoch - 5ms/step
Epoch 456/500
72/72 - 0s - loss: 0.5036 - accuracy: 0.8529 - val_loss: 0.3369 - val_accuracy: 0.9083 - 335ms/epoch - 5ms/step
Epoch 457/500
72/72 - 0s - loss: 0.5074 - accuracy: 0.8491 - val_loss: 0.3486 - val_accuracy: 0.9033 - 331ms/epoch - 5ms/step
Epoch 458/500
72/72 - 0s - loss: 0.4946 - accuracy: 0.8506 - val_loss: 0.3211 - val_accuracy: 0.9123 - 334ms/epoch - 5ms/step
Epoch 459/500
72/72 - 0s - loss: 0.5131 - accuracy: 0.8493 - val_loss: 0.3611 - val_accuracy: 0.8993 - 336ms/epoch - 5ms/step
Epoch 460/500
72/72 - 0s - loss: 0.4907 - accuracy: 0.8519 - val_loss: 0.3910 - val_accuracy: 0.8937 - 336ms/epoch - 5ms/step
Epoch 461/500
72/72 - 0s - loss: 0.4968 - accuracy: 0.8491 - val_loss: 0.3307 - val_accuracy: 0.9080 - 336ms/epoch - 5ms/step
Epoch 462/500
72/72 - 0s - loss: 0.5010 - accuracy: 0.8497 - val_loss: 0.3287 - val_accuracy: 0.9107 - 341ms/epoch - 5ms/step
Epoch 463/500
72/72 - 0s - loss: 0.5178 - accuracy: 0.8434 - val_loss: 0.3365 - val_accuracy: 0.9033 - 341ms/epoch - 5ms/step
Epoch 464/500
72/72 - 0s - loss: 0.5022 - accuracy: 0.8511 - val_loss: 0.3561 - val_accuracy: 0.9020 - 348ms/epoch - 5ms/step
Epoch 465/500
72/72 - 0s - loss: 0.4911 - accuracy: 0.8510 - val_loss: 0.3247 - val_accuracy: 0.9077 - 340ms/epoch - 5ms/step
Epoch 466/500
72/72 - 0s - loss: 0.4982 - accuracy: 0.8564 - val_loss: 0.3295 - val_accuracy: 0.9067 - 335ms/epoch - 5ms/step
Epoch 467/500
72/72 - 0s - loss: 0.5065 - accuracy: 0.8468 - val_loss: 0.3295 - val_accuracy: 0.9080 - 334ms/epoch - 5ms/step
Epoch 468/500
72/72 - 0s - loss: 0.5195 - accuracy: 0.8417 - val_loss: 0.3313 - val_accuracy: 0.9110 - 332ms/epoch - 5ms/step
Epoch 469/500
72/72 - 0s - loss: 0.5045 - accuracy: 0.8473 - val_loss: 0.3649 - val_accuracy: 0.8963 - 343ms/epoch - 5ms/step
Epoch 470/500
72/72 - 0s - loss: 0.5163 - accuracy: 0.8451 - val_loss: 0.3658 - val_accuracy: 0.8960 - 344ms/epoch - 5ms/step
Epoch 471/500
72/72 - 0s - loss: 0.5011 - accuracy: 0.8492 - val_loss: 0.3865 - val_accuracy: 0.8923 - 353ms/epoch - 5ms/step
Epoch 472/500
72/72 - 0s - loss: 0.4943 - accuracy: 0.8498 - val_loss: 0.3388 - val_accuracy: 0.9030 - 353ms/epoch - 5ms/step
Epoch 473/500
72/72 - 0s - loss: 0.5028 - accuracy: 0.8468 - val_loss: 0.3158 - val_accuracy: 0.9117 - 351ms/epoch - 5ms/step
Epoch 474/500
72/72 - 0s - loss: 0.4908 - accuracy: 0.8537 - val_loss: 0.3534 - val_accuracy: 0.9043 - 334ms/epoch - 5ms/step
Epoch 475/500
72/72 - 0s - loss: 0.4906 - accuracy: 0.8512 - val_loss: 0.3330 - val_accuracy: 0.9103 - 339ms/epoch - 5ms/step
Epoch 476/500
72/72 - 0s - loss: 0.5028 - accuracy: 0.8480 - val_loss: 0.3570 - val_accuracy: 0.8993 - 334ms/epoch - 5ms/step
Epoch 477/500
72/72 - 0s - loss: 0.4985 - accuracy: 0.8523 - val_loss: 0.3799 - val_accuracy: 0.8963 - 330ms/epoch - 5ms/step
Epoch 478/500
72/72 - 0s - loss: 0.5088 - accuracy: 0.8505 - val_loss: 0.3423 - val_accuracy: 0.9060 - 331ms/epoch - 5ms/step
Epoch 479/500
72/72 - 0s - loss: 0.5036 - accuracy: 0.8465 - val_loss: 0.3572 - val_accuracy: 0.9003 - 341ms/epoch - 5ms/step
Epoch 480/500
72/72 - 0s - loss: 0.5024 - accuracy: 0.8493 - val_loss: 0.3336 - val_accuracy: 0.9050 - 330ms/epoch - 5ms/step
Epoch 481/500
72/72 - 0s - loss: 0.5048 - accuracy: 0.8488 - val_loss: 0.3337 - val_accuracy: 0.9077 - 337ms/epoch - 5ms/step
Epoch 482/500
72/72 - 0s - loss: 0.4901 - accuracy: 0.8481 - val_loss: 0.3681 - val_accuracy: 0.9000 - 332ms/epoch - 5ms/step
Epoch 483/500
72/72 - 0s - loss: 0.4894 - accuracy: 0.8517 - val_loss: 0.3548 - val_accuracy: 0.9023 - 334ms/epoch - 5ms/step
Epoch 484/500
72/72 - 0s - loss: 0.4929 - accuracy: 0.8510 - val_loss: 0.3314 - val_accuracy: 0.9093 - 338ms/epoch - 5ms/step
Epoch 485/500
72/72 - 0s - loss: 0.5060 - accuracy: 0.8495 - val_loss: 0.3212 - val_accuracy: 0.9090 - 333ms/epoch - 5ms/step
Epoch 486/500
72/72 - 0s - loss: 0.5005 - accuracy: 0.8533 - val_loss: 0.3299 - val_accuracy: 0.9110 - 330ms/epoch - 5ms/step
Epoch 487/500
72/72 - 0s - loss: 0.4920 - accuracy: 0.8511 - val_loss: 0.3399 - val_accuracy: 0.9060 - 333ms/epoch - 5ms/step
Epoch 488/500
72/72 - 0s - loss: 0.5058 - accuracy: 0.8498 - val_loss: 0.3771 - val_accuracy: 0.8950 - 334ms/epoch - 5ms/step
Epoch 489/500
72/72 - 0s - loss: 0.4895 - accuracy: 0.8555 - val_loss: 0.3204 - val_accuracy: 0.9167 - 338ms/epoch - 5ms/step
Epoch 490/500
72/72 - 0s - loss: 0.4857 - accuracy: 0.8538 - val_loss: 0.3270 - val_accuracy: 0.9080 - 338ms/epoch - 5ms/step
Epoch 491/500
72/72 - 0s - loss: 0.4907 - accuracy: 0.8526 - val_loss: 0.3619 - val_accuracy: 0.8993 - 345ms/epoch - 5ms/step
Epoch 492/500
72/72 - 0s - loss: 0.4803 - accuracy: 0.8568 - val_loss: 0.3284 - val_accuracy: 0.9093 - 349ms/epoch - 5ms/step
Epoch 493/500
72/72 - 0s - loss: 0.4898 - accuracy: 0.8510 - val_loss: 0.3562 - val_accuracy: 0.9010 - 340ms/epoch - 5ms/step
Epoch 494/500
72/72 - 0s - loss: 0.5065 - accuracy: 0.8457 - val_loss: 0.3462 - val_accuracy: 0.9043 - 334ms/epoch - 5ms/step
Epoch 495/500
72/72 - 0s - loss: 0.4863 - accuracy: 0.8530 - val_loss: 0.3346 - val_accuracy: 0.9067 - 332ms/epoch - 5ms/step
Epoch 496/500
72/72 - 0s - loss: 0.4928 - accuracy: 0.8513 - val_loss: 0.3517 - val_accuracy: 0.8983 - 329ms/epoch - 5ms/step
Epoch 497/500
72/72 - 0s - loss: 0.4858 - accuracy: 0.8546 - val_loss: 0.3175 - val_accuracy: 0.9123 - 327ms/epoch - 5ms/step
Epoch 498/500
72/72 - 0s - loss: 0.4799 - accuracy: 0.8576 - val_loss: 0.3241 - val_accuracy: 0.9147 - 330ms/epoch - 5ms/step
Epoch 499/500
72/72 - 0s - loss: 0.4939 - accuracy: 0.8510 - val_loss: 0.3314 - val_accuracy: 0.9090 - 327ms/epoch - 5ms/step
Epoch 500/500
72/72 - 0s - loss: 0.5008 - accuracy: 0.8512 - val_loss: 0.3271 - val_accuracy: 0.9087 - 328ms/epoch - 5ms/step
94/94 - 0s - loss: 0.3271 - accuracy: 0.9087 - 329ms/epoch - 3ms/step
Baseline Error: 9.13%

df31augmentationcompare

	Model Name	Train Accuracy	test Accuracy	Val Loss	Val Loss	History
0	With Class Weights	0.854120	0.873000	0.475335	0.482661	<keras.callbacks.History object at 0x7f5a147cb...
1	With Augmented Engineered Data	0.851239	0.908667	0.500761	0.327107	<keras.callbacks.History object at 0x7f594325f...

Model Comparison and Interpretation

Model Evaluation

Model: With Class Weights

• Train Accuracy: 86.48%
• Test Accuracy: 88.13%
• Train Loss: 0.4518
• Test Loss: 0.4229

Model: With Augmented Engineered Data

• Train Accuracy: 87.02%
• Test Accuracy: 90.57%
• Train Loss: 0.4462
• Test Loss: 0.3332

Comparison:

• Accuracy: With Augmented Engineered Data has a higher accuracy on both training and test sets compared to With Class Weights.
• Loss: With Augmented Engineered Data has lower losses on both training and test sets compared to With Class Weights.

Conclusion:

With Augmented Engineered Data outperforms With Class Weights in terms of accuracy and loss on both training and test sets. It appears to be a more effective model based on these metrics.

#save the new "best" model

model.save('Goat31Model.h5')

#to be e xtra safe we can use the test data as well just to make sure

augmentmodel = models.load_model('Goat31Model.h5')

#evaluate the model on the test set
score = augmentmodel.evaluate(X_test31, test_labels, verbose=0)
print("Test loss:", score[0])
print("Test accuracy:", score[1])

Test loss: 0.32786089181900024
Test accuracy: 0.9136666655540466

Model: Regularized Model

• Test Loss: 0.3756
• Test Accuracy: 89.99%

Model: Batch Normalized Model

• Test Loss: 0.4186
• Test Accuracy: 89.43%

Model: Augment Model

• Test Accuracy: 91.40%
• Test Loss: 0.3463

Conclusion:

The Augment Model has the highest accuracy as well as the lowest test loss out of all the other potential "best" models. This is likely due to the data augmentation being effective, one thing we should note is that the augmented model does take slightly longer to converge fully compared to the other models.

Model seems to train better off of the Augmented data, and it reflects on the accuracy when tested on the test data. We should note that the model does take longer to converge which could also imply that the increased accuracy/better loss could be a result of the increase in train data size instead of the balancing being better than class weights

128x128 Model improvements

unlike the 31x31 model, the 128x128 model has more data to work with as the images are more complex, we will have more space to work with as we can add more layers to the models. However we will try to implement the same techniques as we did with the 31x31 model and see if there are any improvements. If you have already forgotten, the initial additions for the 31x31 model are as follows:

• l2 regularization
• increase in epochs

Hopefully this will allow the model to fully converge and learn the data, while preventing overfitting.

#original model

# model_128 = models.Sequential([
#     layers.Conv2D(32, (3, 3), activation='relu', input_shape=(128, 128, 1)),
#     layers.MaxPooling2D((2, 2)),
#     layers.Conv2D(64, (3, 3), activation='relu'),
#     layers.MaxPooling2D((2, 2)),
#     layers.Conv2D(64, (3, 3), activation='relu'),
#     layers.Flatten(),
#     layers.Dense(64, activation='relu'),
#     layers.Dense(15, activation='softmax')
# ])



model_128regularized = models.Sequential([
    layers.Conv2D(32, (3, 3), activation='relu', input_shape=(128, 128, 1)),
    layers.MaxPooling2D((2, 2)),
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.MaxPooling2D((2, 2)),
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.Flatten(),
    layers.Dense(64, activation='relu', kernel_regularizer=regularizers.l2(0.001)), #add reg
    layers.Dense(15, activation='softmax')
])

model_128regularized.compile(optimizer='adam',
              loss= 'categorical_crossentropy',
              metrics=['accuracy'])
history = model_128regularized.fit(X_train128, train_labels, validation_data=(X_val128, validation_labels),
                    epochs=100, batch_size=32, verbose=2, class_weight = class_weight) #add epochs


scores = model_128regularized.evaluate(X_val128, validation_labels, verbose=2)
print("Baseline Error: %.2f%%" % (100-scores[1]*100))
plot_history(history)
df128.loc[len(df128)] = ['RegularizedModel', history.history['accuracy'][-1], history.history['val_accuracy'][-1], history.history['loss'][-1], history.history['val_loss'][-1], history]

Epoch 1/100
283/283 - 4s - loss: 2.5538 - accuracy: 0.2790 - val_loss: 2.0602 - val_accuracy: 0.3870 - 4s/epoch - 15ms/step
Epoch 2/100
283/283 - 2s - loss: 1.7480 - accuracy: 0.5305 - val_loss: 1.6805 - val_accuracy: 0.5227 - 2s/epoch - 7ms/step
Epoch 3/100
283/283 - 2s - loss: 1.3068 - accuracy: 0.6648 - val_loss: 1.4288 - val_accuracy: 0.6210 - 2s/epoch - 7ms/step
Epoch 4/100
283/283 - 2s - loss: 1.0143 - accuracy: 0.7576 - val_loss: 1.2674 - val_accuracy: 0.6910 - 2s/epoch - 7ms/step
Epoch 5/100
283/283 - 2s - loss: 0.8400 - accuracy: 0.8248 - val_loss: 1.3225 - val_accuracy: 0.6980 - 2s/epoch - 7ms/step
Epoch 6/100
283/283 - 2s - loss: 0.6999 - accuracy: 0.8638 - val_loss: 1.2777 - val_accuracy: 0.7170 - 2s/epoch - 7ms/step
Epoch 7/100
283/283 - 2s - loss: 0.5963 - accuracy: 0.9066 - val_loss: 1.2821 - val_accuracy: 0.7347 - 2s/epoch - 7ms/step
Epoch 8/100
283/283 - 2s - loss: 0.5215 - accuracy: 0.9257 - val_loss: 1.5631 - val_accuracy: 0.6697 - 2s/epoch - 7ms/step
Epoch 9/100
283/283 - 2s - loss: 0.5480 - accuracy: 0.9220 - val_loss: 1.2904 - val_accuracy: 0.7500 - 2s/epoch - 7ms/step
Epoch 10/100
283/283 - 2s - loss: 0.5201 - accuracy: 0.9319 - val_loss: 1.4601 - val_accuracy: 0.7230 - 2s/epoch - 7ms/step
Epoch 11/100
283/283 - 2s - loss: 0.5228 - accuracy: 0.9359 - val_loss: 1.5287 - val_accuracy: 0.7307 - 2s/epoch - 7ms/step
Epoch 12/100
283/283 - 2s - loss: 0.4293 - accuracy: 0.9648 - val_loss: 1.4915 - val_accuracy: 0.7357 - 2s/epoch - 7ms/step
Epoch 13/100
283/283 - 2s - loss: 0.4765 - accuracy: 0.9481 - val_loss: 1.3625 - val_accuracy: 0.7560 - 2s/epoch - 7ms/step
Epoch 14/100
283/283 - 2s - loss: 0.4062 - accuracy: 0.9670 - val_loss: 1.4288 - val_accuracy: 0.7610 - 2s/epoch - 7ms/step
Epoch 15/100
283/283 - 2s - loss: 0.4940 - accuracy: 0.9422 - val_loss: 1.5420 - val_accuracy: 0.7347 - 2s/epoch - 7ms/step
Epoch 16/100
283/283 - 2s - loss: 0.4413 - accuracy: 0.9634 - val_loss: 1.4767 - val_accuracy: 0.7543 - 2s/epoch - 7ms/step
Epoch 17/100
283/283 - 2s - loss: 0.4189 - accuracy: 0.9679 - val_loss: 1.4947 - val_accuracy: 0.7603 - 2s/epoch - 7ms/step
Epoch 18/100
283/283 - 2s - loss: 0.4102 - accuracy: 0.9684 - val_loss: 1.7956 - val_accuracy: 0.6977 - 2s/epoch - 7ms/step
Epoch 19/100
283/283 - 2s - loss: 0.4197 - accuracy: 0.9646 - val_loss: 2.1524 - val_accuracy: 0.6247 - 2s/epoch - 7ms/step
Epoch 20/100
283/283 - 2s - loss: 0.3724 - accuracy: 0.9763 - val_loss: 1.5943 - val_accuracy: 0.7467 - 2s/epoch - 7ms/step
Epoch 21/100
283/283 - 2s - loss: 0.4009 - accuracy: 0.9613 - val_loss: 1.7093 - val_accuracy: 0.7303 - 2s/epoch - 7ms/step
Epoch 22/100
283/283 - 2s - loss: 0.4843 - accuracy: 0.9503 - val_loss: 1.5897 - val_accuracy: 0.7340 - 2s/epoch - 7ms/step
Epoch 23/100
283/283 - 2s - loss: 0.3886 - accuracy: 0.9765 - val_loss: 1.7632 - val_accuracy: 0.7200 - 2s/epoch - 7ms/step
Epoch 24/100
283/283 - 2s - loss: 0.3477 - accuracy: 0.9838 - val_loss: 1.5582 - val_accuracy: 0.7530 - 2s/epoch - 7ms/step
Epoch 25/100
283/283 - 2s - loss: 0.3439 - accuracy: 0.9778 - val_loss: 1.6767 - val_accuracy: 0.7283 - 2s/epoch - 7ms/step
Epoch 26/100
283/283 - 2s - loss: 0.4595 - accuracy: 0.9569 - val_loss: 1.5120 - val_accuracy: 0.7577 - 2s/epoch - 7ms/step
Epoch 27/100
283/283 - 2s - loss: 0.3989 - accuracy: 0.9744 - val_loss: 1.7129 - val_accuracy: 0.7280 - 2s/epoch - 7ms/step
Epoch 28/100
283/283 - 2s - loss: 0.3663 - accuracy: 0.9783 - val_loss: 1.6114 - val_accuracy: 0.7450 - 2s/epoch - 7ms/step
Epoch 29/100
283/283 - 2s - loss: 0.3692 - accuracy: 0.9755 - val_loss: 1.5429 - val_accuracy: 0.7593 - 2s/epoch - 7ms/step
Epoch 30/100
283/283 - 2s - loss: 0.4075 - accuracy: 0.9671 - val_loss: 1.8357 - val_accuracy: 0.7307 - 2s/epoch - 7ms/step
Epoch 31/100
283/283 - 2s - loss: 0.3913 - accuracy: 0.9733 - val_loss: 1.5254 - val_accuracy: 0.7630 - 2s/epoch - 7ms/step
Epoch 32/100
283/283 - 2s - loss: 0.3565 - accuracy: 0.9809 - val_loss: 1.8618 - val_accuracy: 0.7220 - 2s/epoch - 7ms/step
Epoch 33/100
283/283 - 2s - loss: 0.4424 - accuracy: 0.9550 - val_loss: 2.0990 - val_accuracy: 0.6710 - 2s/epoch - 7ms/step
Epoch 34/100
283/283 - 2s - loss: 0.3865 - accuracy: 0.9765 - val_loss: 1.6468 - val_accuracy: 0.7473 - 2s/epoch - 7ms/step
Epoch 35/100
283/283 - 2s - loss: 0.3191 - accuracy: 0.9879 - val_loss: 1.6404 - val_accuracy: 0.7453 - 2s/epoch - 7ms/step
Epoch 36/100
283/283 - 2s - loss: 0.2787 - accuracy: 0.9905 - val_loss: 1.5754 - val_accuracy: 0.7390 - 2s/epoch - 7ms/step
Epoch 37/100
283/283 - 2s - loss: 0.3129 - accuracy: 0.9781 - val_loss: 1.5563 - val_accuracy: 0.7270 - 2s/epoch - 7ms/step
Epoch 38/100
283/283 - 2s - loss: 0.4608 - accuracy: 0.9531 - val_loss: 1.7024 - val_accuracy: 0.7327 - 2s/epoch - 7ms/step
Epoch 39/100
283/283 - 2s - loss: 0.3701 - accuracy: 0.9817 - val_loss: 1.8737 - val_accuracy: 0.7273 - 2s/epoch - 7ms/step
Epoch 40/100
283/283 - 2s - loss: 0.3906 - accuracy: 0.9736 - val_loss: 1.6673 - val_accuracy: 0.7510 - 2s/epoch - 7ms/step
Epoch 41/100
283/283 - 2s - loss: 0.3783 - accuracy: 0.9776 - val_loss: 1.6834 - val_accuracy: 0.7470 - 2s/epoch - 7ms/step
Epoch 42/100
283/283 - 2s - loss: 0.3561 - accuracy: 0.9821 - val_loss: 2.0030 - val_accuracy: 0.7037 - 2s/epoch - 7ms/step
Epoch 43/100
283/283 - 2s - loss: 0.3984 - accuracy: 0.9710 - val_loss: 1.8070 - val_accuracy: 0.7317 - 2s/epoch - 7ms/step
Epoch 44/100
283/283 - 2s - loss: 0.3400 - accuracy: 0.9852 - val_loss: 1.7015 - val_accuracy: 0.7363 - 2s/epoch - 7ms/step
Epoch 45/100
283/283 - 2s - loss: 0.3524 - accuracy: 0.9815 - val_loss: 1.8167 - val_accuracy: 0.7353 - 2s/epoch - 7ms/step
Epoch 46/100
283/283 - 2s - loss: 0.3569 - accuracy: 0.9786 - val_loss: 1.6337 - val_accuracy: 0.7483 - 2s/epoch - 7ms/step
Epoch 47/100
283/283 - 2s - loss: 0.3355 - accuracy: 0.9844 - val_loss: 1.7786 - val_accuracy: 0.7363 - 2s/epoch - 7ms/step
Epoch 48/100
283/283 - 2s - loss: 0.2927 - accuracy: 0.9903 - val_loss: 1.7774 - val_accuracy: 0.7423 - 2s/epoch - 7ms/step
Epoch 49/100
283/283 - 2s - loss: 0.3535 - accuracy: 0.9733 - val_loss: 1.8876 - val_accuracy: 0.7210 - 2s/epoch - 7ms/step
Epoch 50/100
283/283 - 2s - loss: 0.4318 - accuracy: 0.9615 - val_loss: 1.8714 - val_accuracy: 0.7220 - 2s/epoch - 7ms/step
Epoch 51/100
283/283 - 2s - loss: 0.3914 - accuracy: 0.9784 - val_loss: 2.0926 - val_accuracy: 0.6940 - 2s/epoch - 7ms/step
Epoch 52/100
283/283 - 2s - loss: 0.3154 - accuracy: 0.9931 - val_loss: 1.5481 - val_accuracy: 0.7770 - 2s/epoch - 7ms/step
Epoch 53/100
283/283 - 2s - loss: 0.2677 - accuracy: 0.9941 - val_loss: 1.5110 - val_accuracy: 0.7793 - 2s/epoch - 7ms/step
Epoch 54/100
283/283 - 2s - loss: 0.2848 - accuracy: 0.9824 - val_loss: 1.7618 - val_accuracy: 0.7183 - 2s/epoch - 7ms/step
Epoch 55/100
283/283 - 2s - loss: 0.4601 - accuracy: 0.9536 - val_loss: 1.9629 - val_accuracy: 0.6907 - 2s/epoch - 7ms/step
Epoch 56/100
283/283 - 2s - loss: 0.3918 - accuracy: 0.9792 - val_loss: 1.6380 - val_accuracy: 0.7483 - 2s/epoch - 7ms/step
Epoch 57/100
283/283 - 2s - loss: 0.3063 - accuracy: 0.9945 - val_loss: 1.4562 - val_accuracy: 0.7727 - 2s/epoch - 7ms/step
Epoch 58/100
283/283 - 2s - loss: 0.2462 - accuracy: 0.9980 - val_loss: 1.4502 - val_accuracy: 0.7747 - 2s/epoch - 7ms/step
Epoch 59/100
283/283 - 2s - loss: 0.1905 - accuracy: 0.9999 - val_loss: 1.4266 - val_accuracy: 0.7677 - 2s/epoch - 7ms/step
Epoch 60/100
283/283 - 2s - loss: 0.1435 - accuracy: 1.0000 - val_loss: 1.2846 - val_accuracy: 0.7780 - 2s/epoch - 7ms/step
Epoch 61/100
283/283 - 2s - loss: 0.1087 - accuracy: 0.9992 - val_loss: 1.4433 - val_accuracy: 0.7393 - 2s/epoch - 7ms/step
Epoch 62/100
283/283 - 2s - loss: 0.6833 - accuracy: 0.8775 - val_loss: 1.5163 - val_accuracy: 0.7167 - 2s/epoch - 7ms/step
Epoch 63/100
283/283 - 2s - loss: 0.4547 - accuracy: 0.9660 - val_loss: 1.4187 - val_accuracy: 0.7687 - 2s/epoch - 7ms/step
Epoch 64/100
283/283 - 2s - loss: 0.3146 - accuracy: 0.9965 - val_loss: 1.4310 - val_accuracy: 0.7753 - 2s/epoch - 7ms/step
Epoch 65/100
283/283 - 2s - loss: 0.2447 - accuracy: 0.9990 - val_loss: 1.3303 - val_accuracy: 0.7807 - 2s/epoch - 7ms/step
Epoch 66/100
283/283 - 2s - loss: 0.1888 - accuracy: 0.9993 - val_loss: 1.2104 - val_accuracy: 0.7840 - 2s/epoch - 7ms/step
Epoch 67/100
283/283 - 2s - loss: 0.3656 - accuracy: 0.9556 - val_loss: 1.7838 - val_accuracy: 0.7027 - 2s/epoch - 7ms/step
Epoch 68/100
283/283 - 2s - loss: 0.4680 - accuracy: 0.9545 - val_loss: 1.6436 - val_accuracy: 0.7347 - 2s/epoch - 7ms/step
Epoch 69/100
283/283 - 2s - loss: 0.3533 - accuracy: 0.9886 - val_loss: 1.6535 - val_accuracy: 0.7510 - 2s/epoch - 7ms/step
Epoch 70/100
283/283 - 2s - loss: 0.2777 - accuracy: 0.9960 - val_loss: 1.5948 - val_accuracy: 0.7540 - 2s/epoch - 7ms/step
Epoch 71/100
283/283 - 2s - loss: 0.2433 - accuracy: 0.9959 - val_loss: 1.4672 - val_accuracy: 0.7653 - 2s/epoch - 7ms/step
Epoch 72/100
283/283 - 2s - loss: 0.2299 - accuracy: 0.9934 - val_loss: 1.7893 - val_accuracy: 0.7180 - 2s/epoch - 7ms/step
Epoch 73/100
283/283 - 2s - loss: 0.4392 - accuracy: 0.9473 - val_loss: 1.6937 - val_accuracy: 0.7237 - 2s/epoch - 7ms/step
Epoch 74/100
283/283 - 2s - loss: 0.4106 - accuracy: 0.9720 - val_loss: 1.5821 - val_accuracy: 0.7597 - 2s/epoch - 7ms/step
Epoch 75/100
283/283 - 2s - loss: 0.3083 - accuracy: 0.9946 - val_loss: 1.5158 - val_accuracy: 0.7683 - 2s/epoch - 7ms/step
Epoch 76/100
283/283 - 2s - loss: 0.2480 - accuracy: 0.9984 - val_loss: 1.3818 - val_accuracy: 0.7880 - 2s/epoch - 7ms/step
Epoch 77/100
283/283 - 2s - loss: 0.1891 - accuracy: 1.0000 - val_loss: 1.3014 - val_accuracy: 0.7910 - 2s/epoch - 7ms/step
Epoch 78/100
283/283 - 2s - loss: 0.1420 - accuracy: 1.0000 - val_loss: 1.2110 - val_accuracy: 0.7943 - 2s/epoch - 7ms/step
Epoch 79/100
283/283 - 2s - loss: 0.1053 - accuracy: 1.0000 - val_loss: 1.1123 - val_accuracy: 0.7970 - 2s/epoch - 7ms/step
Epoch 80/100
283/283 - 2s - loss: 0.0774 - accuracy: 1.0000 - val_loss: 1.0459 - val_accuracy: 0.8013 - 2s/epoch - 7ms/step
Epoch 81/100
283/283 - 2s - loss: 0.1011 - accuracy: 0.9895 - val_loss: 2.2270 - val_accuracy: 0.5737 - 2s/epoch - 7ms/step
Epoch 82/100
283/283 - 2s - loss: 0.9133 - accuracy: 0.8274 - val_loss: 1.4562 - val_accuracy: 0.7383 - 2s/epoch - 7ms/step
Epoch 83/100
283/283 - 2s - loss: 0.4566 - accuracy: 0.9754 - val_loss: 1.3503 - val_accuracy: 0.7843 - 2s/epoch - 7ms/step
Epoch 84/100
283/283 - 2s - loss: 0.3554 - accuracy: 0.9901 - val_loss: 1.4814 - val_accuracy: 0.7647 - 2s/epoch - 7ms/step
Epoch 85/100
283/283 - 2s - loss: 0.2623 - accuracy: 0.9994 - val_loss: 1.2711 - val_accuracy: 0.7903 - 2s/epoch - 7ms/step
Epoch 86/100
283/283 - 2s - loss: 0.2001 - accuracy: 1.0000 - val_loss: 1.1852 - val_accuracy: 0.7930 - 2s/epoch - 7ms/step
Epoch 87/100
283/283 - 2s - loss: 0.1506 - accuracy: 1.0000 - val_loss: 1.0966 - val_accuracy: 0.7933 - 2s/epoch - 7ms/step
Epoch 88/100
283/283 - 2s - loss: 0.1123 - accuracy: 1.0000 - val_loss: 1.0259 - val_accuracy: 0.7907 - 2s/epoch - 7ms/step
Epoch 89/100
283/283 - 2s - loss: 0.0831 - accuracy: 1.0000 - val_loss: 0.9778 - val_accuracy: 0.7923 - 2s/epoch - 7ms/step
Epoch 90/100
283/283 - 2s - loss: 0.1184 - accuracy: 0.9844 - val_loss: 2.0744 - val_accuracy: 0.5840 - 2s/epoch - 7ms/step
Epoch 91/100
283/283 - 2s - loss: 0.8158 - accuracy: 0.8548 - val_loss: 1.3378 - val_accuracy: 0.7597 - 2s/epoch - 7ms/step
Epoch 92/100
283/283 - 2s - loss: 0.4367 - accuracy: 0.9776 - val_loss: 1.2897 - val_accuracy: 0.7787 - 2s/epoch - 7ms/step
Epoch 93/100
283/283 - 2s - loss: 0.3038 - accuracy: 0.9997 - val_loss: 1.1891 - val_accuracy: 0.7957 - 2s/epoch - 7ms/step
Epoch 94/100
283/283 - 2s - loss: 0.2289 - accuracy: 1.0000 - val_loss: 1.1012 - val_accuracy: 0.8003 - 2s/epoch - 7ms/step
Epoch 95/100
283/283 - 2s - loss: 0.1732 - accuracy: 0.9999 - val_loss: 1.0459 - val_accuracy: 0.8003 - 2s/epoch - 7ms/step
Epoch 96/100
283/283 - 2s - loss: 0.1298 - accuracy: 1.0000 - val_loss: 0.9825 - val_accuracy: 0.8050 - 2s/epoch - 7ms/step
Epoch 97/100
283/283 - 2s - loss: 0.0958 - accuracy: 1.0000 - val_loss: 0.9453 - val_accuracy: 0.8040 - 2s/epoch - 7ms/step
Epoch 98/100
283/283 - 2s - loss: 0.0709 - accuracy: 1.0000 - val_loss: 0.9669 - val_accuracy: 0.8000 - 2s/epoch - 7ms/step
Epoch 99/100
283/283 - 2s - loss: 0.6797 - accuracy: 0.8695 - val_loss: 1.2755 - val_accuracy: 0.7703 - 2s/epoch - 7ms/step
Epoch 100/100
283/283 - 2s - loss: 0.4781 - accuracy: 0.9644 - val_loss: 1.5424 - val_accuracy: 0.7543 - 2s/epoch - 7ms/step
94/94 - 0s - loss: 1.5424 - accuracy: 0.7543 - 307ms/epoch - 3ms/step
Baseline Error: 24.57%

As we can see the model is over fitting, there also appears to be fluctuations in the accuracy and loss, this is probably due to the fact that the model is not learning well/struggling to learn the data. We can try to add more layers to the model as well as more regularization(dropout layer) to see if we can improve the performance of the model.

model_128morelayers = models.Sequential([
    layers.Conv2D(32, (3, 3), activation='relu', input_shape=(128, 128, 1)),
    layers.MaxPooling2D((2, 2)),
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.MaxPooling2D((2, 2)),
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.Flatten(),
    layers.Dense(64, activation='relu', kernel_regularizer=regularizers.l2(0.001)),
    layers.Dropout(0.5),  # Adding dropout for regularization
    layers.Dense(32, activation='relu'),  # Adding an additional dense layer
    layers.Dense(15, activation='softmax')
])

model_128morelayers.compile(optimizer='adam',
                loss= 'categorical_crossentropy',
                metrics=['accuracy'])

history = model_128morelayers.fit(X_train128, train_labels, validation_data=(X_val128, validation_labels),
                    epochs=100, batch_size=32, verbose=2, class_weight = class_weight)


scores = model_128morelayers.evaluate(X_val128, validation_labels, verbose=2)
print("Baseline Error: %.2f%%" % (100-scores[1]*100))
plot_history(history)
df128.loc[len(df128)] = ['ExtraThickModel', history.history['accuracy'][-1], history.history['val_accuracy'][-1], history.history['loss'][-1], history.history['val_loss'][-1], history]

Epoch 1/100
283/283 - 5s - loss: 2.8812 - accuracy: 0.1724 - val_loss: 2.5358 - val_accuracy: 0.1723 - 5s/epoch - 18ms/step
Epoch 2/100
283/283 - 2s - loss: 2.4787 - accuracy: 0.2782 - val_loss: 2.2795 - val_accuracy: 0.3067 - 2s/epoch - 7ms/step
Epoch 3/100
283/283 - 2s - loss: 2.2416 - accuracy: 0.3489 - val_loss: 2.2402 - val_accuracy: 0.3203 - 2s/epoch - 7ms/step
Epoch 4/100
283/283 - 2s - loss: 2.1282 - accuracy: 0.3761 - val_loss: 1.9390 - val_accuracy: 0.4270 - 2s/epoch - 7ms/step
Epoch 5/100
283/283 - 2s - loss: 1.9855 - accuracy: 0.4219 - val_loss: 1.8612 - val_accuracy: 0.4387 - 2s/epoch - 7ms/step
Epoch 6/100
283/283 - 2s - loss: 1.9109 - accuracy: 0.4478 - val_loss: 1.7171 - val_accuracy: 0.5153 - 2s/epoch - 7ms/step
Epoch 7/100
283/283 - 2s - loss: 1.8270 - accuracy: 0.4791 - val_loss: 1.7413 - val_accuracy: 0.5180 - 2s/epoch - 7ms/step
Epoch 8/100
283/283 - 2s - loss: 1.7690 - accuracy: 0.4952 - val_loss: 1.6935 - val_accuracy: 0.5317 - 2s/epoch - 7ms/step
Epoch 9/100
283/283 - 2s - loss: 1.7440 - accuracy: 0.5076 - val_loss: 1.7159 - val_accuracy: 0.5380 - 2s/epoch - 7ms/step
Epoch 10/100
283/283 - 2s - loss: 1.6999 - accuracy: 0.5213 - val_loss: 1.6452 - val_accuracy: 0.5610 - 2s/epoch - 7ms/step
Epoch 11/100
283/283 - 2s - loss: 1.6367 - accuracy: 0.5364 - val_loss: 1.5994 - val_accuracy: 0.5787 - 2s/epoch - 7ms/step
Epoch 12/100
283/283 - 2s - loss: 1.6073 - accuracy: 0.5523 - val_loss: 1.4841 - val_accuracy: 0.6373 - 2s/epoch - 7ms/step
Epoch 13/100
283/283 - 2s - loss: 1.5634 - accuracy: 0.5576 - val_loss: 1.5270 - val_accuracy: 0.6117 - 2s/epoch - 7ms/step
Epoch 14/100
283/283 - 2s - loss: 1.5300 - accuracy: 0.5852 - val_loss: 1.5233 - val_accuracy: 0.6127 - 2s/epoch - 7ms/step
Epoch 15/100
283/283 - 2s - loss: 1.4897 - accuracy: 0.6072 - val_loss: 1.4396 - val_accuracy: 0.6283 - 2s/epoch - 7ms/step
Epoch 16/100
283/283 - 2s - loss: 1.4313 - accuracy: 0.6239 - val_loss: 1.3497 - val_accuracy: 0.6760 - 2s/epoch - 7ms/step
Epoch 17/100
283/283 - 2s - loss: 1.4241 - accuracy: 0.6317 - val_loss: 1.4287 - val_accuracy: 0.6553 - 2s/epoch - 7ms/step
Epoch 18/100
283/283 - 2s - loss: 1.4760 - accuracy: 0.6220 - val_loss: 1.6727 - val_accuracy: 0.5920 - 2s/epoch - 7ms/step
Epoch 19/100
283/283 - 2s - loss: 1.3986 - accuracy: 0.6450 - val_loss: 1.4882 - val_accuracy: 0.6423 - 2s/epoch - 7ms/step
Epoch 20/100
283/283 - 2s - loss: 1.3667 - accuracy: 0.6622 - val_loss: 1.4221 - val_accuracy: 0.6627 - 2s/epoch - 7ms/step
Epoch 21/100
283/283 - 2s - loss: 1.3556 - accuracy: 0.6608 - val_loss: 1.3771 - val_accuracy: 0.6900 - 2s/epoch - 7ms/step
Epoch 22/100
283/283 - 2s - loss: 1.3493 - accuracy: 0.6652 - val_loss: 1.3120 - val_accuracy: 0.7040 - 2s/epoch - 7ms/step
Epoch 23/100
283/283 - 2s - loss: 1.2901 - accuracy: 0.6865 - val_loss: 1.8983 - val_accuracy: 0.5920 - 2s/epoch - 7ms/step
Epoch 24/100
283/283 - 2s - loss: 1.3166 - accuracy: 0.6750 - val_loss: 1.4053 - val_accuracy: 0.6753 - 2s/epoch - 7ms/step
Epoch 25/100
283/283 - 2s - loss: 1.2819 - accuracy: 0.6882 - val_loss: 1.3242 - val_accuracy: 0.7040 - 2s/epoch - 7ms/step
Epoch 26/100
283/283 - 2s - loss: 1.2667 - accuracy: 0.6909 - val_loss: 1.3624 - val_accuracy: 0.6963 - 2s/epoch - 7ms/step
Epoch 27/100
283/283 - 2s - loss: 1.2746 - accuracy: 0.6977 - val_loss: 1.3518 - val_accuracy: 0.7043 - 2s/epoch - 7ms/step
Epoch 28/100
283/283 - 2s - loss: 1.2817 - accuracy: 0.6962 - val_loss: 1.4332 - val_accuracy: 0.6763 - 2s/epoch - 7ms/step
Epoch 29/100
283/283 - 2s - loss: 1.2595 - accuracy: 0.7000 - val_loss: 1.2784 - val_accuracy: 0.7247 - 2s/epoch - 7ms/step
Epoch 30/100
283/283 - 2s - loss: 1.2211 - accuracy: 0.7128 - val_loss: 1.3938 - val_accuracy: 0.7000 - 2s/epoch - 7ms/step
Epoch 31/100
283/283 - 2s - loss: 1.2211 - accuracy: 0.7096 - val_loss: 1.2992 - val_accuracy: 0.7230 - 2s/epoch - 7ms/step
Epoch 32/100
283/283 - 2s - loss: 1.2341 - accuracy: 0.7106 - val_loss: 1.3767 - val_accuracy: 0.7040 - 2s/epoch - 7ms/step
Epoch 33/100
283/283 - 2s - loss: 1.2164 - accuracy: 0.7193 - val_loss: 1.2766 - val_accuracy: 0.7310 - 2s/epoch - 7ms/step
Epoch 34/100
283/283 - 2s - loss: 1.2291 - accuracy: 0.7201 - val_loss: 1.2923 - val_accuracy: 0.7310 - 2s/epoch - 7ms/step
Epoch 35/100
283/283 - 2s - loss: 1.2125 - accuracy: 0.7240 - val_loss: 1.3019 - val_accuracy: 0.7247 - 2s/epoch - 7ms/step
Epoch 36/100
283/283 - 2s - loss: 1.1558 - accuracy: 0.7400 - val_loss: 1.2375 - val_accuracy: 0.7390 - 2s/epoch - 7ms/step
Epoch 37/100
283/283 - 2s - loss: 1.1755 - accuracy: 0.7262 - val_loss: 1.3448 - val_accuracy: 0.7087 - 2s/epoch - 7ms/step
Epoch 38/100
283/283 - 2s - loss: 1.1336 - accuracy: 0.7416 - val_loss: 1.3107 - val_accuracy: 0.7307 - 2s/epoch - 7ms/step
Epoch 39/100
283/283 - 2s - loss: 1.1212 - accuracy: 0.7475 - val_loss: 1.4536 - val_accuracy: 0.7073 - 2s/epoch - 7ms/step
Epoch 40/100
283/283 - 2s - loss: 1.1303 - accuracy: 0.7510 - val_loss: 1.4158 - val_accuracy: 0.7000 - 2s/epoch - 7ms/step
Epoch 41/100
283/283 - 2s - loss: 1.1674 - accuracy: 0.7418 - val_loss: 1.2782 - val_accuracy: 0.7310 - 2s/epoch - 7ms/step
Epoch 42/100
283/283 - 2s - loss: 1.1271 - accuracy: 0.7533 - val_loss: 1.2449 - val_accuracy: 0.7413 - 2s/epoch - 7ms/step
Epoch 43/100
283/283 - 2s - loss: 1.1372 - accuracy: 0.7466 - val_loss: 1.3069 - val_accuracy: 0.7287 - 2s/epoch - 7ms/step
Epoch 44/100
283/283 - 2s - loss: 1.1448 - accuracy: 0.7531 - val_loss: 1.2168 - val_accuracy: 0.7553 - 2s/epoch - 7ms/step
Epoch 45/100
283/283 - 2s - loss: 1.1347 - accuracy: 0.7481 - val_loss: 1.3444 - val_accuracy: 0.7267 - 2s/epoch - 7ms/step
Epoch 46/100
283/283 - 2s - loss: 1.1125 - accuracy: 0.7589 - val_loss: 1.3016 - val_accuracy: 0.7397 - 2s/epoch - 7ms/step
Epoch 47/100
283/283 - 2s - loss: 1.1062 - accuracy: 0.7646 - val_loss: 1.3437 - val_accuracy: 0.7133 - 2s/epoch - 7ms/step
Epoch 48/100
283/283 - 2s - loss: 1.0991 - accuracy: 0.7621 - val_loss: 1.2747 - val_accuracy: 0.7430 - 2s/epoch - 7ms/step
Epoch 49/100
283/283 - 2s - loss: 1.1294 - accuracy: 0.7579 - val_loss: 1.3053 - val_accuracy: 0.7337 - 2s/epoch - 7ms/step
Epoch 50/100
283/283 - 2s - loss: 1.0958 - accuracy: 0.7645 - val_loss: 1.3442 - val_accuracy: 0.7237 - 2s/epoch - 7ms/step
Epoch 51/100
283/283 - 2s - loss: 1.1110 - accuracy: 0.7643 - val_loss: 1.2870 - val_accuracy: 0.7357 - 2s/epoch - 7ms/step
Epoch 52/100
283/283 - 2s - loss: 1.1106 - accuracy: 0.7644 - val_loss: 1.2795 - val_accuracy: 0.7377 - 2s/epoch - 7ms/step
Epoch 53/100
283/283 - 2s - loss: 1.0574 - accuracy: 0.7812 - val_loss: 1.3595 - val_accuracy: 0.7197 - 2s/epoch - 7ms/step
Epoch 54/100
283/283 - 2s - loss: 1.0900 - accuracy: 0.7737 - val_loss: 1.2612 - val_accuracy: 0.7460 - 2s/epoch - 7ms/step
Epoch 55/100
283/283 - 2s - loss: 1.1099 - accuracy: 0.7674 - val_loss: 1.2790 - val_accuracy: 0.7403 - 2s/epoch - 7ms/step
Epoch 56/100
283/283 - 2s - loss: 1.0830 - accuracy: 0.7765 - val_loss: 1.2839 - val_accuracy: 0.7470 - 2s/epoch - 7ms/step
Epoch 57/100
283/283 - 2s - loss: 1.0845 - accuracy: 0.7726 - val_loss: 1.3319 - val_accuracy: 0.7243 - 2s/epoch - 7ms/step
Epoch 58/100
283/283 - 2s - loss: 1.0325 - accuracy: 0.7950 - val_loss: 1.4014 - val_accuracy: 0.6960 - 2s/epoch - 7ms/step
Epoch 59/100
283/283 - 2s - loss: 1.0385 - accuracy: 0.7887 - val_loss: 1.3204 - val_accuracy: 0.7287 - 2s/epoch - 7ms/step
Epoch 60/100
283/283 - 2s - loss: 1.0746 - accuracy: 0.7739 - val_loss: 1.2846 - val_accuracy: 0.7350 - 2s/epoch - 7ms/step
Epoch 61/100
283/283 - 2s - loss: 1.0243 - accuracy: 0.7925 - val_loss: 1.2439 - val_accuracy: 0.7560 - 2s/epoch - 7ms/step
Epoch 62/100
283/283 - 2s - loss: 1.0192 - accuracy: 0.7926 - val_loss: 1.2250 - val_accuracy: 0.7513 - 2s/epoch - 7ms/step
Epoch 63/100
283/283 - 2s - loss: 1.0753 - accuracy: 0.7860 - val_loss: 1.2827 - val_accuracy: 0.7417 - 2s/epoch - 7ms/step
Epoch 64/100
283/283 - 2s - loss: 1.0093 - accuracy: 0.7982 - val_loss: 1.3686 - val_accuracy: 0.7397 - 2s/epoch - 7ms/step
Epoch 65/100
283/283 - 2s - loss: 0.9781 - accuracy: 0.8024 - val_loss: 1.1733 - val_accuracy: 0.7717 - 2s/epoch - 7ms/step
Epoch 66/100
283/283 - 2s - loss: 1.0329 - accuracy: 0.7818 - val_loss: 1.2330 - val_accuracy: 0.7667 - 2s/epoch - 7ms/step
Epoch 67/100
283/283 - 2s - loss: 1.0508 - accuracy: 0.7861 - val_loss: 1.3475 - val_accuracy: 0.7323 - 2s/epoch - 7ms/step
Epoch 68/100
283/283 - 2s - loss: 1.0528 - accuracy: 0.7920 - val_loss: 1.3012 - val_accuracy: 0.7377 - 2s/epoch - 7ms/step
Epoch 69/100
283/283 - 2s - loss: 1.0302 - accuracy: 0.7940 - val_loss: 1.2168 - val_accuracy: 0.7567 - 2s/epoch - 7ms/step
Epoch 70/100
283/283 - 2s - loss: 1.0356 - accuracy: 0.7967 - val_loss: 1.3727 - val_accuracy: 0.7350 - 2s/epoch - 7ms/step
Epoch 71/100
283/283 - 2s - loss: 1.0024 - accuracy: 0.8048 - val_loss: 1.2964 - val_accuracy: 0.7567 - 2s/epoch - 7ms/step
Epoch 72/100
283/283 - 2s - loss: 1.0288 - accuracy: 0.7980 - val_loss: 1.2903 - val_accuracy: 0.7480 - 2s/epoch - 7ms/step
Epoch 73/100
283/283 - 2s - loss: 1.0176 - accuracy: 0.8010 - val_loss: 1.3121 - val_accuracy: 0.7463 - 2s/epoch - 7ms/step
Epoch 74/100
283/283 - 2s - loss: 1.0200 - accuracy: 0.7997 - val_loss: 1.2900 - val_accuracy: 0.7390 - 2s/epoch - 7ms/step
Epoch 75/100
283/283 - 2s - loss: 1.0054 - accuracy: 0.8075 - val_loss: 1.3499 - val_accuracy: 0.7280 - 2s/epoch - 7ms/step
Epoch 76/100
283/283 - 2s - loss: 0.9677 - accuracy: 0.8179 - val_loss: 1.2507 - val_accuracy: 0.7623 - 2s/epoch - 8ms/step
Epoch 77/100
283/283 - 2s - loss: 1.0181 - accuracy: 0.7996 - val_loss: 1.2429 - val_accuracy: 0.7630 - 2s/epoch - 7ms/step
Epoch 78/100
283/283 - 2s - loss: 0.9746 - accuracy: 0.8142 - val_loss: 1.2015 - val_accuracy: 0.7850 - 2s/epoch - 7ms/step
Epoch 79/100
283/283 - 2s - loss: 0.9677 - accuracy: 0.8138 - val_loss: 1.1978 - val_accuracy: 0.7717 - 2s/epoch - 7ms/step
Epoch 80/100
283/283 - 2s - loss: 0.9753 - accuracy: 0.8111 - val_loss: 1.1857 - val_accuracy: 0.7797 - 2s/epoch - 7ms/step
Epoch 81/100
283/283 - 2s - loss: 1.0129 - accuracy: 0.8093 - val_loss: 1.2875 - val_accuracy: 0.7507 - 2s/epoch - 7ms/step
Epoch 82/100
283/283 - 2s - loss: 0.9563 - accuracy: 0.8227 - val_loss: 1.1861 - val_accuracy: 0.7800 - 2s/epoch - 7ms/step
Epoch 83/100
283/283 - 2s - loss: 0.9591 - accuracy: 0.8195 - val_loss: 1.2460 - val_accuracy: 0.7643 - 2s/epoch - 7ms/step
Epoch 84/100
283/283 - 2s - loss: 0.9772 - accuracy: 0.8229 - val_loss: 1.1985 - val_accuracy: 0.7690 - 2s/epoch - 7ms/step
Epoch 85/100
283/283 - 2s - loss: 0.9291 - accuracy: 0.8289 - val_loss: 1.1762 - val_accuracy: 0.7747 - 2s/epoch - 7ms/step
Epoch 86/100
283/283 - 2s - loss: 0.9829 - accuracy: 0.8200 - val_loss: 1.2546 - val_accuracy: 0.7577 - 2s/epoch - 7ms/step
Epoch 87/100
283/283 - 2s - loss: 0.9719 - accuracy: 0.8209 - val_loss: 1.2163 - val_accuracy: 0.7710 - 2s/epoch - 7ms/step
Epoch 88/100
283/283 - 2s - loss: 0.9678 - accuracy: 0.8257 - val_loss: 1.1792 - val_accuracy: 0.7673 - 2s/epoch - 7ms/step
Epoch 89/100
283/283 - 2s - loss: 0.9723 - accuracy: 0.8171 - val_loss: 1.2565 - val_accuracy: 0.7567 - 2s/epoch - 7ms/step
Epoch 90/100
283/283 - 2s - loss: 0.9637 - accuracy: 0.8207 - val_loss: 1.4324 - val_accuracy: 0.7133 - 2s/epoch - 7ms/step
Epoch 91/100
283/283 - 2s - loss: 0.9565 - accuracy: 0.8213 - val_loss: 1.3810 - val_accuracy: 0.7383 - 2s/epoch - 7ms/step
Epoch 92/100
283/283 - 2s - loss: 0.9429 - accuracy: 0.8295 - val_loss: 1.2372 - val_accuracy: 0.7737 - 2s/epoch - 7ms/step
Epoch 93/100
283/283 - 2s - loss: 0.9665 - accuracy: 0.8237 - val_loss: 1.1843 - val_accuracy: 0.7800 - 2s/epoch - 7ms/step
Epoch 94/100
283/283 - 2s - loss: 0.9309 - accuracy: 0.8303 - val_loss: 1.1748 - val_accuracy: 0.7833 - 2s/epoch - 7ms/step
Epoch 95/100
283/283 - 2s - loss: 0.9385 - accuracy: 0.8299 - val_loss: 1.1716 - val_accuracy: 0.7843 - 2s/epoch - 7ms/step
Epoch 96/100
283/283 - 2s - loss: 0.9225 - accuracy: 0.8377 - val_loss: 1.1325 - val_accuracy: 0.7893 - 2s/epoch - 7ms/step
Epoch 97/100
283/283 - 2s - loss: 0.9380 - accuracy: 0.8262 - val_loss: 1.2151 - val_accuracy: 0.7713 - 2s/epoch - 7ms/step
Epoch 98/100
283/283 - 2s - loss: 0.9294 - accuracy: 0.8312 - val_loss: 1.2327 - val_accuracy: 0.7673 - 2s/epoch - 7ms/step
Epoch 99/100
283/283 - 2s - loss: 0.9403 - accuracy: 0.8299 - val_loss: 1.1838 - val_accuracy: 0.7850 - 2s/epoch - 7ms/step
Epoch 100/100
283/283 - 2s - loss: 0.9208 - accuracy: 0.8353 - val_loss: 1.2471 - val_accuracy: 0.7660 - 2s/epoch - 7ms/step
94/94 - 0s - loss: 1.2471 - accuracy: 0.7660 - 302ms/epoch - 3ms/step
Baseline Error: 23.40%

We can see that the model still seems to be overfitting and there is still some fluctuations, we should first try to stabilize the model by adding more regularization layers or through other means like image augmentation or applying batch normalization. As the model training is slightly slow, i think that applying batch normalization would be a good idea as it would allow the model to converge faster as well as making the model more stable, which will solve 2 of my problems at once.

model_128BN = models.Sequential([
    layers.Conv2D(32, (3, 3), activation='relu', input_shape=(128, 128, 1)),
    layers.MaxPooling2D((2, 2)),
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.MaxPooling2D((2, 2)),
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.Flatten(),
    layers.Dense(64, activation='relu', kernel_regularizer=regularizers.l2(0.001)),
    layers.Dropout(0.5),
    layers.Dense(32, activation='relu'),
    layers.BatchNormalization(), #add batch normalization
    layers.Dense(15, activation='softmax')
])

model_128BN.compile(optimizer='adam',
                loss= 'categorical_crossentropy',
                metrics=['accuracy'])

history = model_128BN.fit(X_train128, train_labels, validation_data=(X_val128, validation_labels),
                    epochs=100, batch_size=32, verbose=2, class_weight = class_weight)


scores = model_128BN.evaluate(X_val128, validation_labels, verbose=2)
print("Baseline Error: %.2f%%" % (100-scores[1]*100))
plot_history(history)
df128.loc[len(df128)] = ['BNModel', history.history['accuracy'][-1], history.history['val_accuracy'][-1], history.history['loss'][-1], history.history['val_loss'][-1], history]

Epoch 1/100
283/283 - 5s - loss: 3.4392 - accuracy: 0.1115 - val_loss: 2.7890 - val_accuracy: 0.0853 - 5s/epoch - 18ms/step
Epoch 2/100
283/283 - 2s - loss: 2.9918 - accuracy: 0.1653 - val_loss: 2.9038 - val_accuracy: 0.1007 - 2s/epoch - 8ms/step
Epoch 3/100
283/283 - 2s - loss: 2.7965 - accuracy: 0.2067 - val_loss: 2.8479 - val_accuracy: 0.1143 - 2s/epoch - 8ms/step
Epoch 4/100
283/283 - 2s - loss: 2.6494 - accuracy: 0.2547 - val_loss: 2.7457 - val_accuracy: 0.1447 - 2s/epoch - 8ms/step
Epoch 5/100
283/283 - 2s - loss: 2.5035 - accuracy: 0.2994 - val_loss: 2.5735 - val_accuracy: 0.2213 - 2s/epoch - 8ms/step
Epoch 6/100
283/283 - 2s - loss: 2.5430 - accuracy: 0.2953 - val_loss: 2.3594 - val_accuracy: 0.3077 - 2s/epoch - 8ms/step
Epoch 7/100
283/283 - 2s - loss: 2.3426 - accuracy: 0.3477 - val_loss: 2.4077 - val_accuracy: 0.2917 - 2s/epoch - 8ms/step
Epoch 8/100
283/283 - 2s - loss: 2.3587 - accuracy: 0.3510 - val_loss: 2.1295 - val_accuracy: 0.3817 - 2s/epoch - 8ms/step
Epoch 9/100
283/283 - 2s - loss: 2.1896 - accuracy: 0.3969 - val_loss: 2.0155 - val_accuracy: 0.4313 - 2s/epoch - 8ms/step
Epoch 10/100
283/283 - 2s - loss: 2.1031 - accuracy: 0.4228 - val_loss: 1.8990 - val_accuracy: 0.4683 - 2s/epoch - 8ms/step
Epoch 11/100
283/283 - 2s - loss: 2.0431 - accuracy: 0.4435 - val_loss: 1.8384 - val_accuracy: 0.5103 - 2s/epoch - 8ms/step
Epoch 12/100
283/283 - 2s - loss: 1.9108 - accuracy: 0.4745 - val_loss: 3.8159 - val_accuracy: 0.1300 - 2s/epoch - 8ms/step
Epoch 13/100
283/283 - 2s - loss: 2.7060 - accuracy: 0.3086 - val_loss: 2.1527 - val_accuracy: 0.3710 - 2s/epoch - 8ms/step
Epoch 14/100
283/283 - 2s - loss: 1.8885 - accuracy: 0.4796 - val_loss: 1.7734 - val_accuracy: 0.5163 - 2s/epoch - 8ms/step
Epoch 15/100
283/283 - 2s - loss: 1.7798 - accuracy: 0.5183 - val_loss: 1.6346 - val_accuracy: 0.5720 - 2s/epoch - 8ms/step
Epoch 16/100
283/283 - 2s - loss: 1.7144 - accuracy: 0.5402 - val_loss: 1.6883 - val_accuracy: 0.5520 - 2s/epoch - 8ms/step
Epoch 17/100
283/283 - 2s - loss: 1.6559 - accuracy: 0.5575 - val_loss: 1.9102 - val_accuracy: 0.4860 - 2s/epoch - 8ms/step
Epoch 18/100
283/283 - 2s - loss: 1.9823 - accuracy: 0.4983 - val_loss: 1.7499 - val_accuracy: 0.5483 - 2s/epoch - 7ms/step
Epoch 19/100
283/283 - 2s - loss: 1.7148 - accuracy: 0.5579 - val_loss: 1.6632 - val_accuracy: 0.5790 - 2s/epoch - 7ms/step
Epoch 20/100
283/283 - 2s - loss: 1.5819 - accuracy: 0.5920 - val_loss: 1.6773 - val_accuracy: 0.5733 - 2s/epoch - 8ms/step
Epoch 21/100
283/283 - 2s - loss: 1.5613 - accuracy: 0.5974 - val_loss: 1.5473 - val_accuracy: 0.6247 - 2s/epoch - 8ms/step
Epoch 22/100
283/283 - 2s - loss: 1.5153 - accuracy: 0.6182 - val_loss: 1.4335 - val_accuracy: 0.6660 - 2s/epoch - 8ms/step
Epoch 23/100
283/283 - 2s - loss: 1.4870 - accuracy: 0.6348 - val_loss: 1.7947 - val_accuracy: 0.5643 - 2s/epoch - 8ms/step
Epoch 24/100
283/283 - 2s - loss: 1.4552 - accuracy: 0.6461 - val_loss: 1.4867 - val_accuracy: 0.6513 - 2s/epoch - 8ms/step
Epoch 25/100
283/283 - 2s - loss: 1.4396 - accuracy: 0.6522 - val_loss: 1.4843 - val_accuracy: 0.6347 - 2s/epoch - 7ms/step
Epoch 26/100
283/283 - 2s - loss: 1.3835 - accuracy: 0.6678 - val_loss: 1.5298 - val_accuracy: 0.6300 - 2s/epoch - 8ms/step
Epoch 27/100
283/283 - 2s - loss: 1.4232 - accuracy: 0.6678 - val_loss: 1.5109 - val_accuracy: 0.6553 - 2s/epoch - 8ms/step
Epoch 28/100
283/283 - 2s - loss: 1.3484 - accuracy: 0.6830 - val_loss: 1.3453 - val_accuracy: 0.7027 - 2s/epoch - 8ms/step
Epoch 29/100
283/283 - 2s - loss: 1.3287 - accuracy: 0.6872 - val_loss: 1.3401 - val_accuracy: 0.6993 - 2s/epoch - 7ms/step
Epoch 30/100
283/283 - 2s - loss: 1.3908 - accuracy: 0.6767 - val_loss: 1.3759 - val_accuracy: 0.6900 - 2s/epoch - 8ms/step
Epoch 31/100
283/283 - 2s - loss: 1.3719 - accuracy: 0.6838 - val_loss: 1.3424 - val_accuracy: 0.7093 - 2s/epoch - 8ms/step
Epoch 32/100
283/283 - 2s - loss: 1.2868 - accuracy: 0.7026 - val_loss: 1.5283 - val_accuracy: 0.6490 - 2s/epoch - 8ms/step
Epoch 33/100
283/283 - 2s - loss: 1.2681 - accuracy: 0.7127 - val_loss: 1.4256 - val_accuracy: 0.6853 - 2s/epoch - 8ms/step
Epoch 34/100
283/283 - 2s - loss: 1.2565 - accuracy: 0.7133 - val_loss: 1.2794 - val_accuracy: 0.7303 - 2s/epoch - 8ms/step
Epoch 35/100
283/283 - 2s - loss: 1.2551 - accuracy: 0.7144 - val_loss: 1.2962 - val_accuracy: 0.7187 - 2s/epoch - 8ms/step
Epoch 36/100
283/283 - 2s - loss: 1.2669 - accuracy: 0.7126 - val_loss: 1.3423 - val_accuracy: 0.6990 - 2s/epoch - 8ms/step
Epoch 37/100
283/283 - 2s - loss: 1.2046 - accuracy: 0.7346 - val_loss: 1.3755 - val_accuracy: 0.6893 - 2s/epoch - 8ms/step
Epoch 38/100
283/283 - 2s - loss: 1.3182 - accuracy: 0.7125 - val_loss: 1.3744 - val_accuracy: 0.6960 - 2s/epoch - 8ms/step
Epoch 39/100
283/283 - 2s - loss: 1.1917 - accuracy: 0.7436 - val_loss: 1.4559 - val_accuracy: 0.6790 - 2s/epoch - 8ms/step
Epoch 40/100
283/283 - 2s - loss: 1.1963 - accuracy: 0.7388 - val_loss: 1.4839 - val_accuracy: 0.6640 - 2s/epoch - 7ms/step
Epoch 41/100
283/283 - 2s - loss: 1.2504 - accuracy: 0.7240 - val_loss: 1.3604 - val_accuracy: 0.6967 - 2s/epoch - 8ms/step
Epoch 42/100
283/283 - 2s - loss: 1.1978 - accuracy: 0.7434 - val_loss: 1.3946 - val_accuracy: 0.6967 - 2s/epoch - 8ms/step
Epoch 43/100
283/283 - 2s - loss: 1.1824 - accuracy: 0.7459 - val_loss: 1.2944 - val_accuracy: 0.7267 - 2s/epoch - 8ms/step
Epoch 44/100
283/283 - 2s - loss: 1.1504 - accuracy: 0.7506 - val_loss: 1.2656 - val_accuracy: 0.7250 - 2s/epoch - 8ms/step
Epoch 45/100
283/283 - 2s - loss: 1.1801 - accuracy: 0.7473 - val_loss: 1.4329 - val_accuracy: 0.6840 - 2s/epoch - 8ms/step
Epoch 46/100
283/283 - 2s - loss: 1.1830 - accuracy: 0.7488 - val_loss: 1.4045 - val_accuracy: 0.6897 - 2s/epoch - 8ms/step
Epoch 47/100
283/283 - 2s - loss: 1.1408 - accuracy: 0.7636 - val_loss: 1.5179 - val_accuracy: 0.6587 - 2s/epoch - 7ms/step
Epoch 48/100
283/283 - 2s - loss: 1.1188 - accuracy: 0.7654 - val_loss: 1.3120 - val_accuracy: 0.7157 - 2s/epoch - 8ms/step
Epoch 49/100
283/283 - 2s - loss: 1.1039 - accuracy: 0.7698 - val_loss: 1.2970 - val_accuracy: 0.7230 - 2s/epoch - 8ms/step
Epoch 50/100
283/283 - 2s - loss: 1.1265 - accuracy: 0.7634 - val_loss: 1.3948 - val_accuracy: 0.6977 - 2s/epoch - 8ms/step
Epoch 51/100
283/283 - 2s - loss: 1.1139 - accuracy: 0.7668 - val_loss: 1.3010 - val_accuracy: 0.7227 - 2s/epoch - 8ms/step
Epoch 52/100
283/283 - 2s - loss: 1.0966 - accuracy: 0.7750 - val_loss: 1.4001 - val_accuracy: 0.7070 - 2s/epoch - 8ms/step
Epoch 53/100
283/283 - 2s - loss: 1.1044 - accuracy: 0.7734 - val_loss: 1.2972 - val_accuracy: 0.7243 - 2s/epoch - 8ms/step
Epoch 54/100
283/283 - 2s - loss: 1.0858 - accuracy: 0.7760 - val_loss: 1.2901 - val_accuracy: 0.7247 - 2s/epoch - 8ms/step
Epoch 55/100
283/283 - 2s - loss: 1.1199 - accuracy: 0.7655 - val_loss: 1.3006 - val_accuracy: 0.7203 - 2s/epoch - 8ms/step
Epoch 56/100
283/283 - 2s - loss: 1.0705 - accuracy: 0.7792 - val_loss: 1.4242 - val_accuracy: 0.6830 - 2s/epoch - 8ms/step
Epoch 57/100
283/283 - 2s - loss: 1.0561 - accuracy: 0.7905 - val_loss: 1.3399 - val_accuracy: 0.7137 - 2s/epoch - 8ms/step
Epoch 58/100
283/283 - 2s - loss: 1.0536 - accuracy: 0.7879 - val_loss: 1.3857 - val_accuracy: 0.7037 - 2s/epoch - 8ms/step
Epoch 59/100
283/283 - 2s - loss: 1.0456 - accuracy: 0.7936 - val_loss: 1.4647 - val_accuracy: 0.6850 - 2s/epoch - 8ms/step
Epoch 60/100
283/283 - 2s - loss: 1.0817 - accuracy: 0.7781 - val_loss: 1.4635 - val_accuracy: 0.6890 - 2s/epoch - 8ms/step
Epoch 61/100
283/283 - 2s - loss: 1.0725 - accuracy: 0.7852 - val_loss: 1.3276 - val_accuracy: 0.7200 - 2s/epoch - 8ms/step
Epoch 62/100
283/283 - 2s - loss: 1.0565 - accuracy: 0.7844 - val_loss: 1.4217 - val_accuracy: 0.6873 - 2s/epoch - 8ms/step
Epoch 63/100
283/283 - 2s - loss: 1.0297 - accuracy: 0.7980 - val_loss: 1.3568 - val_accuracy: 0.7090 - 2s/epoch - 8ms/step
Epoch 64/100
283/283 - 2s - loss: 1.0931 - accuracy: 0.7775 - val_loss: 1.3791 - val_accuracy: 0.7090 - 2s/epoch - 7ms/step
Epoch 65/100
283/283 - 2s - loss: 1.0986 - accuracy: 0.7800 - val_loss: 1.3623 - val_accuracy: 0.7177 - 2s/epoch - 7ms/step
Epoch 66/100
283/283 - 2s - loss: 1.0409 - accuracy: 0.7966 - val_loss: 1.5560 - val_accuracy: 0.6713 - 2s/epoch - 7ms/step
Epoch 67/100
283/283 - 2s - loss: 1.0493 - accuracy: 0.7918 - val_loss: 1.3573 - val_accuracy: 0.7170 - 2s/epoch - 7ms/step
Epoch 68/100
283/283 - 2s - loss: 1.0252 - accuracy: 0.8041 - val_loss: 1.3173 - val_accuracy: 0.7303 - 2s/epoch - 7ms/step
Epoch 69/100
283/283 - 2s - loss: 0.9857 - accuracy: 0.8064 - val_loss: 1.3156 - val_accuracy: 0.7190 - 2s/epoch - 7ms/step
Epoch 70/100
283/283 - 2s - loss: 1.0402 - accuracy: 0.7963 - val_loss: 1.4080 - val_accuracy: 0.7090 - 2s/epoch - 8ms/step
Epoch 71/100
283/283 - 2s - loss: 1.0307 - accuracy: 0.8028 - val_loss: 1.4405 - val_accuracy: 0.6940 - 2s/epoch - 8ms/step
Epoch 72/100
283/283 - 2s - loss: 1.0050 - accuracy: 0.8070 - val_loss: 1.3891 - val_accuracy: 0.6977 - 2s/epoch - 8ms/step
Epoch 73/100
283/283 - 2s - loss: 1.0434 - accuracy: 0.7985 - val_loss: 1.3774 - val_accuracy: 0.7183 - 2s/epoch - 7ms/step
Epoch 74/100
283/283 - 2s - loss: 1.0326 - accuracy: 0.7909 - val_loss: 1.3687 - val_accuracy: 0.7177 - 2s/epoch - 8ms/step
Epoch 75/100
283/283 - 2s - loss: 1.0077 - accuracy: 0.8055 - val_loss: 1.7147 - val_accuracy: 0.6247 - 2s/epoch - 7ms/step
Epoch 76/100
283/283 - 2s - loss: 0.9917 - accuracy: 0.8130 - val_loss: 1.5604 - val_accuracy: 0.6767 - 2s/epoch - 7ms/step
Epoch 77/100
283/283 - 2s - loss: 1.0537 - accuracy: 0.7979 - val_loss: 1.4223 - val_accuracy: 0.7027 - 2s/epoch - 8ms/step
Epoch 78/100
283/283 - 2s - loss: 0.9821 - accuracy: 0.8138 - val_loss: 1.6633 - val_accuracy: 0.6587 - 2s/epoch - 8ms/step
Epoch 79/100
283/283 - 2s - loss: 1.1394 - accuracy: 0.7798 - val_loss: 1.3772 - val_accuracy: 0.7200 - 2s/epoch - 8ms/step
Epoch 80/100
283/283 - 2s - loss: 1.0464 - accuracy: 0.8017 - val_loss: 1.4488 - val_accuracy: 0.6910 - 2s/epoch - 7ms/step
Epoch 81/100
283/283 - 2s - loss: 0.9613 - accuracy: 0.8161 - val_loss: 1.5798 - val_accuracy: 0.6670 - 2s/epoch - 8ms/step
Epoch 82/100
283/283 - 2s - loss: 0.9806 - accuracy: 0.8201 - val_loss: 1.3902 - val_accuracy: 0.7137 - 2s/epoch - 8ms/step
Epoch 83/100
283/283 - 2s - loss: 1.0113 - accuracy: 0.8129 - val_loss: 1.4386 - val_accuracy: 0.6937 - 2s/epoch - 8ms/step
Epoch 84/100
283/283 - 2s - loss: 1.0360 - accuracy: 0.8073 - val_loss: 1.5480 - val_accuracy: 0.6747 - 2s/epoch - 8ms/step
Epoch 85/100
283/283 - 2s - loss: 0.9516 - accuracy: 0.8330 - val_loss: 1.3873 - val_accuracy: 0.7257 - 2s/epoch - 8ms/step
Epoch 86/100
283/283 - 2s - loss: 0.9885 - accuracy: 0.8158 - val_loss: 1.4499 - val_accuracy: 0.6993 - 2s/epoch - 8ms/step
Epoch 87/100
283/283 - 2s - loss: 1.0081 - accuracy: 0.8114 - val_loss: 1.4305 - val_accuracy: 0.7093 - 2s/epoch - 7ms/step
Epoch 88/100
283/283 - 2s - loss: 0.9829 - accuracy: 0.8190 - val_loss: 1.4197 - val_accuracy: 0.7047 - 2s/epoch - 8ms/step
Epoch 89/100
283/283 - 2s - loss: 0.9615 - accuracy: 0.8222 - val_loss: 1.4087 - val_accuracy: 0.7107 - 2s/epoch - 7ms/step
Epoch 90/100
283/283 - 2s - loss: 0.9623 - accuracy: 0.8178 - val_loss: 1.4780 - val_accuracy: 0.6923 - 2s/epoch - 7ms/step
Epoch 91/100
283/283 - 2s - loss: 0.9611 - accuracy: 0.8158 - val_loss: 1.4663 - val_accuracy: 0.6953 - 2s/epoch - 8ms/step
Epoch 92/100
283/283 - 2s - loss: 0.9713 - accuracy: 0.8158 - val_loss: 1.4091 - val_accuracy: 0.7097 - 2s/epoch - 8ms/step
Epoch 93/100
283/283 - 2s - loss: 0.9364 - accuracy: 0.8327 - val_loss: 1.4059 - val_accuracy: 0.7103 - 2s/epoch - 8ms/step
Epoch 94/100
283/283 - 2s - loss: 0.9288 - accuracy: 0.8269 - val_loss: 1.4002 - val_accuracy: 0.7100 - 2s/epoch - 8ms/step
Epoch 95/100
283/283 - 2s - loss: 0.9441 - accuracy: 0.8247 - val_loss: 1.4027 - val_accuracy: 0.7080 - 2s/epoch - 8ms/step
Epoch 96/100
283/283 - 2s - loss: 0.9350 - accuracy: 0.8324 - val_loss: 1.4092 - val_accuracy: 0.7070 - 2s/epoch - 8ms/step
Epoch 97/100
283/283 - 2s - loss: 0.9386 - accuracy: 0.8281 - val_loss: 1.5582 - val_accuracy: 0.6783 - 2s/epoch - 8ms/step
Epoch 98/100
283/283 - 2s - loss: 0.9675 - accuracy: 0.8200 - val_loss: 1.5581 - val_accuracy: 0.6703 - 2s/epoch - 8ms/step
Epoch 99/100
283/283 - 2s - loss: 0.9214 - accuracy: 0.8335 - val_loss: 1.4141 - val_accuracy: 0.7143 - 2s/epoch - 8ms/step
Epoch 100/100
283/283 - 2s - loss: 0.9467 - accuracy: 0.8261 - val_loss: 1.3638 - val_accuracy: 0.7377 - 2s/epoch - 8ms/step
94/94 - 0s - loss: 1.3638 - accuracy: 0.7377 - 304ms/epoch - 3ms/step
Baseline Error: 26.23%

The graph seems to be slightly better compared to the previous graph, however the accuracy is still bottle necking at around 70%. To solve this issue, we can try to use lr decay in order to let the model learn better, hopefully leading to better results.

model_128BNlrd = models.Sequential([
    layers.Conv2D(32, (3, 3), activation='relu', input_shape=(128, 128, 1)),
    layers.MaxPooling2D((2, 2)),
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.MaxPooling2D((2, 2)),
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.Flatten(),
    layers.Dense(64, activation='relu', kernel_regularizer=regularizers.l2(0.001)),
    layers.Dropout(0.5),
    layers.Dense(32, activation='relu'),
    layers.BatchNormalization(),
    layers.Dense(15, activation='softmax')
])

model_128BNlrd.compile(optimizer='adam',
                loss= 'categorical_crossentropy',
                metrics=['accuracy'])

history = model_128BNlrd.fit(X_train128, train_labels, validation_data=(X_val128, validation_labels),
                    epochs=100, batch_size=32, verbose=2, class_weight = class_weight,
                    callbacks=[ReduceLROnPlateau(monitor='val_loss', factor=0.8, patience=10, verbose=1, mode='auto', min_delta=0.0001, cooldown=0, min_lr=0)]) #lr decay


scores = model_128BNlrd.evaluate(X_val128, validation_labels, verbose=2)
print("Baseline Error: %.2f%%" % (100-scores[1]*100))
plot_history(history)
df128.loc[len(df128)] = ['BNlrdModel', history.history['accuracy'][-1], history.history['val_accuracy'][-1], history.history['loss'][-1], history.history['val_loss'][-1], history]

Epoch 1/100
283/283 - 5s - loss: 3.4319 - accuracy: 0.1107 - val_loss: 2.8222 - val_accuracy: 0.0843 - lr: 0.0010 - 5s/epoch - 18ms/step
Epoch 2/100
283/283 - 2s - loss: 2.9866 - accuracy: 0.1571 - val_loss: 3.0187 - val_accuracy: 0.1297 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 3/100
283/283 - 2s - loss: 2.9096 - accuracy: 0.1661 - val_loss: 2.7145 - val_accuracy: 0.1373 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 4/100
283/283 - 2s - loss: 2.8584 - accuracy: 0.1718 - val_loss: 2.7253 - val_accuracy: 0.1240 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 5/100
283/283 - 2s - loss: 2.7834 - accuracy: 0.1788 - val_loss: 2.6522 - val_accuracy: 0.1443 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 6/100
283/283 - 2s - loss: 2.7047 - accuracy: 0.2209 - val_loss: 2.6407 - val_accuracy: 0.1970 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 7/100
283/283 - 2s - loss: 2.6536 - accuracy: 0.2353 - val_loss: 2.5766 - val_accuracy: 0.1980 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 8/100
283/283 - 2s - loss: 2.5346 - accuracy: 0.2574 - val_loss: 2.5174 - val_accuracy: 0.2423 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 9/100
283/283 - 2s - loss: 2.4348 - accuracy: 0.2821 - val_loss: 2.3946 - val_accuracy: 0.2973 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 10/100
283/283 - 2s - loss: 2.4360 - accuracy: 0.2982 - val_loss: 2.1883 - val_accuracy: 0.3687 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 11/100
283/283 - 2s - loss: 2.2262 - accuracy: 0.3455 - val_loss: 2.1186 - val_accuracy: 0.3723 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 12/100
283/283 - 2s - loss: 2.0845 - accuracy: 0.3872 - val_loss: 2.0613 - val_accuracy: 0.4243 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 13/100
283/283 - 2s - loss: 1.9880 - accuracy: 0.4153 - val_loss: 2.0253 - val_accuracy: 0.4300 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 14/100
283/283 - 2s - loss: 1.9413 - accuracy: 0.4518 - val_loss: 1.7861 - val_accuracy: 0.5273 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 15/100
283/283 - 2s - loss: 1.8648 - accuracy: 0.4821 - val_loss: 1.7109 - val_accuracy: 0.5447 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 16/100
283/283 - 2s - loss: 1.7857 - accuracy: 0.5115 - val_loss: 1.6704 - val_accuracy: 0.5787 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 17/100
283/283 - 2s - loss: 1.7549 - accuracy: 0.5216 - val_loss: 1.6717 - val_accuracy: 0.5820 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 18/100
283/283 - 2s - loss: 1.6627 - accuracy: 0.5513 - val_loss: 1.6423 - val_accuracy: 0.5867 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 19/100
283/283 - 2s - loss: 1.5994 - accuracy: 0.5830 - val_loss: 1.6058 - val_accuracy: 0.5733 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 20/100
283/283 - 2s - loss: 1.6192 - accuracy: 0.5864 - val_loss: 1.6096 - val_accuracy: 0.6187 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 21/100
283/283 - 2s - loss: 1.5874 - accuracy: 0.5970 - val_loss: 1.6689 - val_accuracy: 0.5917 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 22/100
283/283 - 2s - loss: 1.5334 - accuracy: 0.6231 - val_loss: 1.4975 - val_accuracy: 0.6307 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 23/100
283/283 - 2s - loss: 1.4857 - accuracy: 0.6366 - val_loss: 1.4852 - val_accuracy: 0.6387 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 24/100
283/283 - 2s - loss: 1.4810 - accuracy: 0.6376 - val_loss: 1.5422 - val_accuracy: 0.6197 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 25/100
283/283 - 2s - loss: 1.4393 - accuracy: 0.6516 - val_loss: 1.4078 - val_accuracy: 0.6740 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 26/100
283/283 - 2s - loss: 1.4485 - accuracy: 0.6577 - val_loss: 1.6110 - val_accuracy: 0.6110 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 27/100
283/283 - 2s - loss: 1.4163 - accuracy: 0.6802 - val_loss: 1.3862 - val_accuracy: 0.6990 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 28/100
283/283 - 2s - loss: 1.4253 - accuracy: 0.6733 - val_loss: 1.5261 - val_accuracy: 0.6473 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 29/100
283/283 - 2s - loss: 1.3775 - accuracy: 0.6936 - val_loss: 1.3771 - val_accuracy: 0.6980 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 30/100
283/283 - 2s - loss: 1.3616 - accuracy: 0.6845 - val_loss: 1.5702 - val_accuracy: 0.6320 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 31/100
283/283 - 2s - loss: 1.3405 - accuracy: 0.7007 - val_loss: 1.4274 - val_accuracy: 0.6803 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 32/100
283/283 - 2s - loss: 1.3575 - accuracy: 0.6963 - val_loss: 1.4000 - val_accuracy: 0.6960 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 33/100
283/283 - 2s - loss: 1.2972 - accuracy: 0.7126 - val_loss: 1.3674 - val_accuracy: 0.7093 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 34/100
283/283 - 2s - loss: 1.2924 - accuracy: 0.7121 - val_loss: 1.3835 - val_accuracy: 0.7043 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 35/100
283/283 - 2s - loss: 1.2876 - accuracy: 0.7218 - val_loss: 1.3448 - val_accuracy: 0.7223 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 36/100
283/283 - 2s - loss: 1.2804 - accuracy: 0.7253 - val_loss: 1.3186 - val_accuracy: 0.7293 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 37/100
283/283 - 2s - loss: 1.2517 - accuracy: 0.7277 - val_loss: 1.2920 - val_accuracy: 0.7307 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 38/100
283/283 - 2s - loss: 1.2805 - accuracy: 0.7190 - val_loss: 1.3787 - val_accuracy: 0.7133 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 39/100
283/283 - 2s - loss: 1.2854 - accuracy: 0.7189 - val_loss: 1.3080 - val_accuracy: 0.7370 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 40/100
283/283 - 2s - loss: 1.2309 - accuracy: 0.7453 - val_loss: 1.4346 - val_accuracy: 0.6870 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 41/100
283/283 - 2s - loss: 1.2906 - accuracy: 0.7239 - val_loss: 1.3229 - val_accuracy: 0.7390 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 42/100
283/283 - 2s - loss: 1.2364 - accuracy: 0.7389 - val_loss: 1.3021 - val_accuracy: 0.7323 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 43/100
283/283 - 2s - loss: 1.2489 - accuracy: 0.7333 - val_loss: 1.2974 - val_accuracy: 0.7353 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 44/100
283/283 - 2s - loss: 1.2256 - accuracy: 0.7434 - val_loss: 1.3968 - val_accuracy: 0.7003 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 45/100
283/283 - 2s - loss: 1.1840 - accuracy: 0.7482 - val_loss: 1.5325 - val_accuracy: 0.6583 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 46/100
283/283 - 2s - loss: 1.2159 - accuracy: 0.7451 - val_loss: 1.2511 - val_accuracy: 0.7467 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 47/100
283/283 - 2s - loss: 1.2222 - accuracy: 0.7467 - val_loss: 1.3774 - val_accuracy: 0.7033 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 48/100
283/283 - 2s - loss: 1.1817 - accuracy: 0.7549 - val_loss: 1.4124 - val_accuracy: 0.6940 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 49/100
283/283 - 2s - loss: 1.1929 - accuracy: 0.7519 - val_loss: 1.3115 - val_accuracy: 0.7197 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 50/100
283/283 - 2s - loss: 1.2102 - accuracy: 0.7466 - val_loss: 1.3186 - val_accuracy: 0.7223 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 51/100
283/283 - 2s - loss: 1.1864 - accuracy: 0.7571 - val_loss: 1.2531 - val_accuracy: 0.7470 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 52/100
283/283 - 2s - loss: 1.1694 - accuracy: 0.7558 - val_loss: 1.2661 - val_accuracy: 0.7387 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 53/100
283/283 - 2s - loss: 1.1301 - accuracy: 0.7632 - val_loss: 1.4019 - val_accuracy: 0.6910 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 54/100
283/283 - 2s - loss: 1.1961 - accuracy: 0.7544 - val_loss: 1.2769 - val_accuracy: 0.7407 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 55/100
283/283 - 2s - loss: 1.1649 - accuracy: 0.7653 - val_loss: 1.2118 - val_accuracy: 0.7590 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 56/100
283/283 - 2s - loss: 1.1243 - accuracy: 0.7736 - val_loss: 1.3412 - val_accuracy: 0.7213 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 57/100
283/283 - 2s - loss: 1.1314 - accuracy: 0.7687 - val_loss: 1.2862 - val_accuracy: 0.7370 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 58/100
283/283 - 2s - loss: 1.1371 - accuracy: 0.7657 - val_loss: 1.2332 - val_accuracy: 0.7450 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 59/100
283/283 - 2s - loss: 1.1029 - accuracy: 0.7750 - val_loss: 1.2533 - val_accuracy: 0.7477 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 60/100
283/283 - 2s - loss: 1.1416 - accuracy: 0.7706 - val_loss: 1.2427 - val_accuracy: 0.7560 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 61/100
283/283 - 2s - loss: 1.1693 - accuracy: 0.7592 - val_loss: 1.3230 - val_accuracy: 0.7287 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 62/100
283/283 - 2s - loss: 1.1502 - accuracy: 0.7666 - val_loss: 1.2497 - val_accuracy: 0.7610 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 63/100
283/283 - 2s - loss: 1.1184 - accuracy: 0.7753 - val_loss: 1.2191 - val_accuracy: 0.7573 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 64/100
283/283 - 2s - loss: 1.1282 - accuracy: 0.7759 - val_loss: 1.2930 - val_accuracy: 0.7280 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 65/100
283/283 - 2s - loss: 1.0766 - accuracy: 0.7843 - val_loss: 1.2012 - val_accuracy: 0.7707 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 66/100
283/283 - 2s - loss: 1.1015 - accuracy: 0.7774 - val_loss: 1.3396 - val_accuracy: 0.7180 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 67/100
283/283 - 2s - loss: 1.1272 - accuracy: 0.7723 - val_loss: 1.3074 - val_accuracy: 0.7277 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 68/100
283/283 - 2s - loss: 1.1153 - accuracy: 0.7779 - val_loss: 1.1848 - val_accuracy: 0.7777 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 69/100
283/283 - 2s - loss: 1.1324 - accuracy: 0.7777 - val_loss: 1.2679 - val_accuracy: 0.7557 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 70/100
283/283 - 2s - loss: 1.1143 - accuracy: 0.7830 - val_loss: 1.3005 - val_accuracy: 0.7393 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 71/100
283/283 - 2s - loss: 1.1663 - accuracy: 0.7691 - val_loss: 1.3360 - val_accuracy: 0.7360 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 72/100
283/283 - 2s - loss: 1.1183 - accuracy: 0.7858 - val_loss: 1.3498 - val_accuracy: 0.7263 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 73/100
283/283 - 2s - loss: 1.0814 - accuracy: 0.7874 - val_loss: 1.2805 - val_accuracy: 0.7557 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 74/100
283/283 - 2s - loss: 1.1018 - accuracy: 0.7883 - val_loss: 1.3851 - val_accuracy: 0.7090 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 75/100
283/283 - 2s - loss: 1.1600 - accuracy: 0.7687 - val_loss: 1.2906 - val_accuracy: 0.7440 - lr: 0.0010 - 2s/epoch - 7ms/step
Epoch 76/100
283/283 - 2s - loss: 1.0711 - accuracy: 0.8008 - val_loss: 1.3697 - val_accuracy: 0.7083 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 77/100
283/283 - 2s - loss: 1.0655 - accuracy: 0.7970 - val_loss: 1.2958 - val_accuracy: 0.7413 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 78/100

Epoch 78: ReduceLROnPlateau reducing learning rate to 0.000800000037997961.
283/283 - 2s - loss: 1.1027 - accuracy: 0.7844 - val_loss: 1.3624 - val_accuracy: 0.7170 - lr: 0.0010 - 2s/epoch - 8ms/step
Epoch 79/100
283/283 - 2s - loss: 1.0159 - accuracy: 0.8025 - val_loss: 1.2130 - val_accuracy: 0.7543 - lr: 8.0000e-04 - 2s/epoch - 8ms/step
Epoch 80/100
283/283 - 2s - loss: 0.9631 - accuracy: 0.8182 - val_loss: 1.2089 - val_accuracy: 0.7570 - lr: 8.0000e-04 - 2s/epoch - 7ms/step
Epoch 81/100
283/283 - 2s - loss: 0.9503 - accuracy: 0.8154 - val_loss: 1.2013 - val_accuracy: 0.7490 - lr: 8.0000e-04 - 2s/epoch - 7ms/step
Epoch 82/100
283/283 - 2s - loss: 0.9484 - accuracy: 0.8191 - val_loss: 1.1721 - val_accuracy: 0.7590 - lr: 8.0000e-04 - 2s/epoch - 7ms/step
Epoch 83/100
283/283 - 2s - loss: 0.9217 - accuracy: 0.8187 - val_loss: 1.1953 - val_accuracy: 0.7513 - lr: 8.0000e-04 - 2s/epoch - 7ms/step
Epoch 84/100
283/283 - 2s - loss: 0.9369 - accuracy: 0.8131 - val_loss: 1.1944 - val_accuracy: 0.7577 - lr: 8.0000e-04 - 2s/epoch - 8ms/step
Epoch 85/100
283/283 - 2s - loss: 0.9250 - accuracy: 0.8206 - val_loss: 1.3066 - val_accuracy: 0.7267 - lr: 8.0000e-04 - 2s/epoch - 7ms/step
Epoch 86/100
283/283 - 2s - loss: 0.9117 - accuracy: 0.8230 - val_loss: 1.1884 - val_accuracy: 0.7573 - lr: 8.0000e-04 - 2s/epoch - 7ms/step
Epoch 87/100
283/283 - 2s - loss: 0.9324 - accuracy: 0.8176 - val_loss: 1.2777 - val_accuracy: 0.7313 - lr: 8.0000e-04 - 2s/epoch - 8ms/step
Epoch 88/100
283/283 - 2s - loss: 0.9226 - accuracy: 0.8197 - val_loss: 1.2433 - val_accuracy: 0.7360 - lr: 8.0000e-04 - 2s/epoch - 8ms/step
Epoch 89/100
283/283 - 2s - loss: 0.9488 - accuracy: 0.8162 - val_loss: 1.2206 - val_accuracy: 0.7387 - lr: 8.0000e-04 - 2s/epoch - 8ms/step
Epoch 90/100
283/283 - 2s - loss: 0.9304 - accuracy: 0.8183 - val_loss: 1.2002 - val_accuracy: 0.7483 - lr: 8.0000e-04 - 2s/epoch - 8ms/step
Epoch 91/100
283/283 - 2s - loss: 0.9397 - accuracy: 0.8197 - val_loss: 1.2531 - val_accuracy: 0.7433 - lr: 8.0000e-04 - 2s/epoch - 7ms/step
Epoch 92/100

Epoch 92: ReduceLROnPlateau reducing learning rate to 0.0006400000303983689.
283/283 - 2s - loss: 0.9214 - accuracy: 0.8247 - val_loss: 1.2786 - val_accuracy: 0.7397 - lr: 8.0000e-04 - 2s/epoch - 8ms/step
Epoch 93/100
283/283 - 2s - loss: 0.8977 - accuracy: 0.8301 - val_loss: 1.1732 - val_accuracy: 0.7610 - lr: 6.4000e-04 - 2s/epoch - 8ms/step
Epoch 94/100
283/283 - 2s - loss: 0.8353 - accuracy: 0.8443 - val_loss: 1.2760 - val_accuracy: 0.7207 - lr: 6.4000e-04 - 2s/epoch - 7ms/step
Epoch 95/100
283/283 - 2s - loss: 0.7898 - accuracy: 0.8564 - val_loss: 1.1579 - val_accuracy: 0.7603 - lr: 6.4000e-04 - 2s/epoch - 8ms/step
Epoch 96/100
283/283 - 2s - loss: 0.8003 - accuracy: 0.8457 - val_loss: 1.1358 - val_accuracy: 0.7597 - lr: 6.4000e-04 - 2s/epoch - 8ms/step
Epoch 97/100
283/283 - 2s - loss: 0.7865 - accuracy: 0.8490 - val_loss: 1.1218 - val_accuracy: 0.7573 - lr: 6.4000e-04 - 2s/epoch - 8ms/step
Epoch 98/100
283/283 - 2s - loss: 0.7733 - accuracy: 0.8558 - val_loss: 1.1637 - val_accuracy: 0.7570 - lr: 6.4000e-04 - 2s/epoch - 8ms/step
Epoch 99/100
283/283 - 2s - loss: 0.7976 - accuracy: 0.8466 - val_loss: 1.1847 - val_accuracy: 0.7510 - lr: 6.4000e-04 - 2s/epoch - 8ms/step
Epoch 100/100
283/283 - 2s - loss: 0.8170 - accuracy: 0.8440 - val_loss: 1.1840 - val_accuracy: 0.7473 - lr: 6.4000e-04 - 2s/epoch - 8ms/step
94/94 - 0s - loss: 1.1840 - accuracy: 0.7473 - 310ms/epoch - 3ms/step
Baseline Error: 25.27%

From the graph, we can see that the model has stabilized a lot, now we just need to add some regularization layers to fix overfitting issue

model_128BNlrdReg = models.Sequential([
    layers.Conv2D(32, (3, 3), activation='relu', input_shape=(128, 128, 1)),
    layers.MaxPooling2D((2, 2)),
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.MaxPooling2D((2, 2)),
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.Dropout(0.5),
    layers.Flatten(),
    layers.Dense(64, activation='relu', kernel_regularizer=regularizers.l2(0.001)),
    layers.Dropout(0.5),
    layers.Dense(32, activation='relu'),
    layers.BatchNormalization(),
    layers.Dense(15, activation='softmax')
])

model_128BNlrdReg.compile(optimizer='adam',
                loss= 'categorical_crossentropy',
                metrics=['accuracy'])

history = model_128BNlrdReg.fit(X_train128, train_labels, validation_data=(X_val128, validation_labels),
                    epochs=100, batch_size=32, verbose=2, class_weight = class_weight,
                    callbacks=[ReduceLROnPlateau(monitor='val_loss', factor=0.8, patience=10, verbose=1, mode='auto', min_delta=0.0001, cooldown=0, min_lr=0)]) #lr decay


scores = model_128BNlrdReg.evaluate(X_val128, validation_labels, verbose=2)
print("Baseline Error: %.2f%%" % (100-scores[1]*100))
plot_history(history)
df128.loc[len(df128)] = ['BNlrdRegModel', history.history['accuracy'][-1], history.history['val_accuracy'][-1], history.history['loss'][-1], history.history['val_loss'][-1], history]

Epoch 1/100

2023-11-26 13:44:12.486696: E tensorflow/core/grappler/optimizers/meta_optimizer.cc:954] layout failed: INVALID_ARGUMENT: Size of values 0 does not match size of permutation 4 @ fanin shape insequential_18/dropout_22/dropout/SelectV2-2-TransposeNHWCToNCHW-LayoutOptimizer

283/283 - 6s - loss: 3.5422 - accuracy: 0.0886 - val_loss: 2.8525 - val_accuracy: 0.0663 - lr: 0.0010 - 6s/epoch - 20ms/step
Epoch 2/100
283/283 - 3s - loss: 3.1700 - accuracy: 0.1360 - val_loss: 2.9036 - val_accuracy: 0.1030 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 3/100
283/283 - 3s - loss: 3.0094 - accuracy: 0.1611 - val_loss: 2.8424 - val_accuracy: 0.1157 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 4/100
283/283 - 3s - loss: 2.8992 - accuracy: 0.1741 - val_loss: 2.7713 - val_accuracy: 0.1397 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 5/100
283/283 - 3s - loss: 2.8954 - accuracy: 0.1740 - val_loss: 2.8554 - val_accuracy: 0.1200 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 6/100
283/283 - 3s - loss: 2.8303 - accuracy: 0.1873 - val_loss: 2.7977 - val_accuracy: 0.1247 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 7/100
283/283 - 3s - loss: 2.8189 - accuracy: 0.1999 - val_loss: 2.6420 - val_accuracy: 0.1763 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 8/100
283/283 - 3s - loss: 2.7850 - accuracy: 0.2131 - val_loss: 2.6222 - val_accuracy: 0.1847 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 9/100
283/283 - 3s - loss: 2.6856 - accuracy: 0.2434 - val_loss: 2.5768 - val_accuracy: 0.2170 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 10/100
283/283 - 3s - loss: 2.6362 - accuracy: 0.2610 - val_loss: 2.4972 - val_accuracy: 0.2487 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 11/100
283/283 - 3s - loss: 2.5614 - accuracy: 0.2712 - val_loss: 2.4379 - val_accuracy: 0.2583 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 12/100
283/283 - 3s - loss: 2.5106 - accuracy: 0.2861 - val_loss: 2.2692 - val_accuracy: 0.3233 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 13/100
283/283 - 3s - loss: 2.4080 - accuracy: 0.3141 - val_loss: 2.2834 - val_accuracy: 0.3100 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 14/100
283/283 - 3s - loss: 2.3227 - accuracy: 0.3365 - val_loss: 2.3558 - val_accuracy: 0.2957 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 15/100
283/283 - 3s - loss: 2.2967 - accuracy: 0.3492 - val_loss: 2.2657 - val_accuracy: 0.3143 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 16/100
283/283 - 3s - loss: 2.2140 - accuracy: 0.3778 - val_loss: 2.1286 - val_accuracy: 0.3807 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 17/100
283/283 - 2s - loss: 2.1355 - accuracy: 0.4054 - val_loss: 2.0822 - val_accuracy: 0.4057 - lr: 0.0010 - 2s/epoch - 9ms/step
Epoch 18/100
283/283 - 2s - loss: 2.0635 - accuracy: 0.4320 - val_loss: 1.9172 - val_accuracy: 0.4710 - lr: 0.0010 - 2s/epoch - 9ms/step
Epoch 19/100
283/283 - 3s - loss: 1.9885 - accuracy: 0.4578 - val_loss: 2.0107 - val_accuracy: 0.4223 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 20/100
283/283 - 3s - loss: 1.9051 - accuracy: 0.4821 - val_loss: 1.8197 - val_accuracy: 0.5053 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 21/100
283/283 - 3s - loss: 1.8847 - accuracy: 0.5007 - val_loss: 1.7178 - val_accuracy: 0.5510 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 22/100
283/283 - 3s - loss: 1.8577 - accuracy: 0.5167 - val_loss: 1.9005 - val_accuracy: 0.4753 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 23/100
283/283 - 2s - loss: 1.8084 - accuracy: 0.5358 - val_loss: 1.7777 - val_accuracy: 0.5330 - lr: 0.0010 - 2s/epoch - 9ms/step
Epoch 24/100
283/283 - 2s - loss: 1.7245 - accuracy: 0.5636 - val_loss: 1.9676 - val_accuracy: 0.4893 - lr: 0.0010 - 2s/epoch - 9ms/step
Epoch 25/100
283/283 - 2s - loss: 1.8596 - accuracy: 0.5377 - val_loss: 1.6073 - val_accuracy: 0.6013 - lr: 0.0010 - 2s/epoch - 9ms/step
Epoch 26/100
283/283 - 2s - loss: 1.6715 - accuracy: 0.5837 - val_loss: 1.6792 - val_accuracy: 0.5880 - lr: 0.0010 - 2s/epoch - 9ms/step
Epoch 27/100
283/283 - 3s - loss: 1.6363 - accuracy: 0.6010 - val_loss: 1.4992 - val_accuracy: 0.6580 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 28/100
283/283 - 3s - loss: 1.6220 - accuracy: 0.6086 - val_loss: 1.5807 - val_accuracy: 0.6373 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 29/100
283/283 - 3s - loss: 1.5989 - accuracy: 0.6122 - val_loss: 1.4592 - val_accuracy: 0.6723 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 30/100
283/283 - 3s - loss: 1.5712 - accuracy: 0.6277 - val_loss: 1.4957 - val_accuracy: 0.6550 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 31/100
283/283 - 2s - loss: 1.5695 - accuracy: 0.6294 - val_loss: 1.4029 - val_accuracy: 0.6870 - lr: 0.0010 - 2s/epoch - 9ms/step
Epoch 32/100
283/283 - 2s - loss: 1.5218 - accuracy: 0.6391 - val_loss: 1.3725 - val_accuracy: 0.6960 - lr: 0.0010 - 2s/epoch - 9ms/step
Epoch 33/100
283/283 - 2s - loss: 1.4935 - accuracy: 0.6491 - val_loss: 1.3751 - val_accuracy: 0.6953 - lr: 0.0010 - 2s/epoch - 9ms/step
Epoch 34/100
283/283 - 3s - loss: 1.4601 - accuracy: 0.6654 - val_loss: 1.3412 - val_accuracy: 0.7073 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 35/100
283/283 - 2s - loss: 1.4468 - accuracy: 0.6659 - val_loss: 1.5295 - val_accuracy: 0.6530 - lr: 0.0010 - 2s/epoch - 9ms/step
Epoch 36/100
283/283 - 2s - loss: 1.6871 - accuracy: 0.6093 - val_loss: 1.3743 - val_accuracy: 0.7107 - lr: 0.0010 - 2s/epoch - 9ms/step
Epoch 37/100
283/283 - 2s - loss: 1.4717 - accuracy: 0.6687 - val_loss: 1.3845 - val_accuracy: 0.7053 - lr: 0.0010 - 2s/epoch - 9ms/step
Epoch 38/100
283/283 - 2s - loss: 1.4059 - accuracy: 0.6843 - val_loss: 1.3055 - val_accuracy: 0.7217 - lr: 0.0010 - 2s/epoch - 9ms/step
Epoch 39/100
283/283 - 3s - loss: 1.4259 - accuracy: 0.6768 - val_loss: 1.3491 - val_accuracy: 0.7090 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 40/100
283/283 - 3s - loss: 1.4114 - accuracy: 0.6872 - val_loss: 1.3241 - val_accuracy: 0.7070 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 41/100
283/283 - 2s - loss: 1.4005 - accuracy: 0.6883 - val_loss: 1.2371 - val_accuracy: 0.7497 - lr: 0.0010 - 2s/epoch - 9ms/step
Epoch 42/100
283/283 - 3s - loss: 1.3920 - accuracy: 0.6953 - val_loss: 1.2202 - val_accuracy: 0.7670 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 43/100
283/283 - 3s - loss: 1.3651 - accuracy: 0.7000 - val_loss: 1.2836 - val_accuracy: 0.7383 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 44/100
283/283 - 3s - loss: 1.3960 - accuracy: 0.6936 - val_loss: 1.2769 - val_accuracy: 0.7420 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 45/100
283/283 - 3s - loss: 1.3743 - accuracy: 0.7041 - val_loss: 1.2489 - val_accuracy: 0.7503 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 46/100
283/283 - 2s - loss: 1.3373 - accuracy: 0.7082 - val_loss: 1.3167 - val_accuracy: 0.7163 - lr: 0.0010 - 2s/epoch - 9ms/step
Epoch 47/100
283/283 - 2s - loss: 1.3401 - accuracy: 0.7151 - val_loss: 1.2218 - val_accuracy: 0.7660 - lr: 0.0010 - 2s/epoch - 9ms/step
Epoch 48/100
283/283 - 2s - loss: 1.3743 - accuracy: 0.7053 - val_loss: 1.2739 - val_accuracy: 0.7440 - lr: 0.0010 - 2s/epoch - 9ms/step
Epoch 49/100
283/283 - 2s - loss: 1.3376 - accuracy: 0.7095 - val_loss: 1.4213 - val_accuracy: 0.6863 - lr: 0.0010 - 2s/epoch - 9ms/step
Epoch 50/100
283/283 - 2s - loss: 1.3367 - accuracy: 0.7140 - val_loss: 1.1540 - val_accuracy: 0.7867 - lr: 0.0010 - 2s/epoch - 9ms/step
Epoch 51/100
283/283 - 3s - loss: 1.2971 - accuracy: 0.7235 - val_loss: 1.1673 - val_accuracy: 0.7717 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 52/100
283/283 - 3s - loss: 1.3019 - accuracy: 0.7240 - val_loss: 1.3627 - val_accuracy: 0.7137 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 53/100
283/283 - 3s - loss: 1.3305 - accuracy: 0.7198 - val_loss: 1.2260 - val_accuracy: 0.7513 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 54/100
283/283 - 3s - loss: 1.3356 - accuracy: 0.7180 - val_loss: 1.2879 - val_accuracy: 0.7330 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 55/100
283/283 - 3s - loss: 1.2989 - accuracy: 0.7304 - val_loss: 1.1844 - val_accuracy: 0.7733 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 56/100
283/283 - 3s - loss: 1.3015 - accuracy: 0.7357 - val_loss: 1.2467 - val_accuracy: 0.7430 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 57/100
283/283 - 2s - loss: 1.3083 - accuracy: 0.7309 - val_loss: 1.3086 - val_accuracy: 0.7270 - lr: 0.0010 - 2s/epoch - 9ms/step
Epoch 58/100
283/283 - 3s - loss: 1.2787 - accuracy: 0.7405 - val_loss: 1.3301 - val_accuracy: 0.7177 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 59/100
283/283 - 2s - loss: 1.3237 - accuracy: 0.7252 - val_loss: 1.2640 - val_accuracy: 0.7567 - lr: 0.0010 - 2s/epoch - 9ms/step
Epoch 60/100

Epoch 60: ReduceLROnPlateau reducing learning rate to 0.000800000037997961.
283/283 - 2s - loss: 1.4047 - accuracy: 0.7174 - val_loss: 1.1620 - val_accuracy: 0.7920 - lr: 0.0010 - 2s/epoch - 9ms/step
Epoch 61/100
283/283 - 2s - loss: 1.2507 - accuracy: 0.7520 - val_loss: 1.2160 - val_accuracy: 0.7553 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 62/100
283/283 - 2s - loss: 1.2166 - accuracy: 0.7519 - val_loss: 1.1330 - val_accuracy: 0.7737 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 63/100
283/283 - 3s - loss: 1.1532 - accuracy: 0.7704 - val_loss: 1.2495 - val_accuracy: 0.7357 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 64/100
283/283 - 2s - loss: 1.1674 - accuracy: 0.7651 - val_loss: 1.0435 - val_accuracy: 0.8130 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 65/100
283/283 - 3s - loss: 1.1518 - accuracy: 0.7653 - val_loss: 1.1250 - val_accuracy: 0.7783 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 66/100
283/283 - 2s - loss: 1.1349 - accuracy: 0.7684 - val_loss: 1.0568 - val_accuracy: 0.8020 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 67/100
283/283 - 3s - loss: 1.1469 - accuracy: 0.7679 - val_loss: 1.1103 - val_accuracy: 0.7783 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 68/100
283/283 - 2s - loss: 1.1155 - accuracy: 0.7729 - val_loss: 1.0306 - val_accuracy: 0.8063 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 69/100
283/283 - 2s - loss: 1.1467 - accuracy: 0.7623 - val_loss: 1.0582 - val_accuracy: 0.8017 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 70/100
283/283 - 2s - loss: 1.1416 - accuracy: 0.7729 - val_loss: 1.1218 - val_accuracy: 0.7740 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 71/100
283/283 - 2s - loss: 1.1615 - accuracy: 0.7635 - val_loss: 1.0773 - val_accuracy: 0.7903 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 72/100
283/283 - 2s - loss: 1.0868 - accuracy: 0.7883 - val_loss: 1.1262 - val_accuracy: 0.7720 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 73/100
283/283 - 2s - loss: 1.2951 - accuracy: 0.7366 - val_loss: 1.0768 - val_accuracy: 0.7980 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 74/100
283/283 - 2s - loss: 1.1343 - accuracy: 0.7719 - val_loss: 1.1072 - val_accuracy: 0.7940 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 75/100
283/283 - 2s - loss: 1.1278 - accuracy: 0.7714 - val_loss: 1.0631 - val_accuracy: 0.7947 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 76/100
283/283 - 3s - loss: 1.0606 - accuracy: 0.7935 - val_loss: 1.0270 - val_accuracy: 0.8110 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 77/100
283/283 - 3s - loss: 1.0851 - accuracy: 0.7836 - val_loss: 1.0737 - val_accuracy: 0.7857 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 78/100
283/283 - 2s - loss: 1.0729 - accuracy: 0.7903 - val_loss: 1.0875 - val_accuracy: 0.7787 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 79/100
283/283 - 2s - loss: 1.0570 - accuracy: 0.7904 - val_loss: 1.0630 - val_accuracy: 0.7963 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 80/100
283/283 - 2s - loss: 1.0704 - accuracy: 0.7894 - val_loss: 1.1247 - val_accuracy: 0.7663 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 81/100
283/283 - 2s - loss: 1.0742 - accuracy: 0.7874 - val_loss: 1.0333 - val_accuracy: 0.7917 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 82/100
283/283 - 2s - loss: 1.0699 - accuracy: 0.7894 - val_loss: 0.9854 - val_accuracy: 0.8123 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 83/100
283/283 - 2s - loss: 1.0652 - accuracy: 0.7908 - val_loss: 1.0012 - val_accuracy: 0.8123 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 84/100
283/283 - 2s - loss: 1.0644 - accuracy: 0.7938 - val_loss: 0.9689 - val_accuracy: 0.8317 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 85/100
283/283 - 2s - loss: 1.0449 - accuracy: 0.7942 - val_loss: 0.9964 - val_accuracy: 0.8150 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 86/100
283/283 - 3s - loss: 1.0611 - accuracy: 0.7926 - val_loss: 1.0034 - val_accuracy: 0.8093 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 87/100
283/283 - 3s - loss: 1.0536 - accuracy: 0.7945 - val_loss: 1.0575 - val_accuracy: 0.7927 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 88/100
283/283 - 2s - loss: 1.0693 - accuracy: 0.7913 - val_loss: 1.0119 - val_accuracy: 0.8130 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 89/100
283/283 - 2s - loss: 1.0978 - accuracy: 0.7877 - val_loss: 1.0387 - val_accuracy: 0.8023 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 90/100
283/283 - 2s - loss: 1.0937 - accuracy: 0.7858 - val_loss: 0.9855 - val_accuracy: 0.8200 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 91/100
283/283 - 3s - loss: 1.0395 - accuracy: 0.8035 - val_loss: 0.9721 - val_accuracy: 0.8213 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 92/100
283/283 - 3s - loss: 1.0432 - accuracy: 0.7955 - val_loss: 1.0471 - val_accuracy: 0.7977 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 93/100
283/283 - 2s - loss: 1.0299 - accuracy: 0.8100 - val_loss: 1.0614 - val_accuracy: 0.8013 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 94/100

Epoch 94: ReduceLROnPlateau reducing learning rate to 0.0006400000303983689.
283/283 - 2s - loss: 1.0801 - accuracy: 0.7966 - val_loss: 0.9740 - val_accuracy: 0.8270 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 95/100
283/283 - 2s - loss: 1.0064 - accuracy: 0.8148 - val_loss: 0.9737 - val_accuracy: 0.8240 - lr: 6.4000e-04 - 2s/epoch - 9ms/step
Epoch 96/100
283/283 - 2s - loss: 0.9995 - accuracy: 0.8094 - val_loss: 0.9449 - val_accuracy: 0.8310 - lr: 6.4000e-04 - 2s/epoch - 9ms/step
Epoch 97/100
283/283 - 2s - loss: 0.9911 - accuracy: 0.8130 - val_loss: 0.9276 - val_accuracy: 0.8380 - lr: 6.4000e-04 - 2s/epoch - 9ms/step
Epoch 98/100
283/283 - 2s - loss: 0.9544 - accuracy: 0.8219 - val_loss: 0.9804 - val_accuracy: 0.8140 - lr: 6.4000e-04 - 2s/epoch - 9ms/step
Epoch 99/100
283/283 - 2s - loss: 0.9362 - accuracy: 0.8216 - val_loss: 0.9211 - val_accuracy: 0.8277 - lr: 6.4000e-04 - 2s/epoch - 9ms/step
Epoch 100/100
283/283 - 2s - loss: 0.9465 - accuracy: 0.8186 - val_loss: 0.8594 - val_accuracy: 0.8523 - lr: 6.4000e-04 - 2s/epoch - 9ms/step
94/94 - 0s - loss: 0.8594 - accuracy: 0.8523 - 305ms/epoch - 3ms/step
Baseline Error: 14.77%

We can see that the model has excellent performance now, without much overfitting, not much jittering as well as a decent accuracy of almost 90%. The slight jitters are likely caused by the changes in learning rate. However as i will not be doing any tuning, the learning rate portion of the model will not be touched up on. We can see that the model has not fully converged, as the model already has a decent performance, i believe that letting it fully converge will result in an even better model with excellenet performance.

model_128BNlrdReg = models.Sequential([
    layers.Conv2D(32, (3, 3), activation='relu', input_shape=(128, 128, 1)),
    layers.MaxPooling2D((2, 2)),
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.MaxPooling2D((2, 2)),
    layers.Conv2D(64, (3, 3), activation='relu'),
    layers.Dropout(0.5),
    layers.Flatten(),
    layers.Dense(64, activation='relu', kernel_regularizer=regularizers.l2(0.001)),
    layers.Dropout(0.5),
    layers.Dense(32, activation='relu'),
    layers.BatchNormalization(),
    layers.Dense(15, activation='softmax')
])

model_128BNlrdReg.compile(optimizer='adam',
                loss= 'categorical_crossentropy',
                metrics=['accuracy'])

history = model_128BNlrdReg.fit(X_train128, train_labels, validation_data=(X_val128, validation_labels),
                    epochs=350, batch_size=32, verbose=2, class_weight = class_weight, #increase epochs for guaranteed convergence
                    callbacks=[ReduceLROnPlateau(monitor='val_loss', factor=0.8, patience=10, verbose=1, mode='auto', min_delta=0.0001, cooldown=0, min_lr=0)]) #lr decay


scores = model_128BNlrdReg.evaluate(X_val128, validation_labels, verbose=2)
print("Baseline Error: %.2f%%" % (100-scores[1]*100))
plot_history(history)
df128.loc[len(df128)] = ['BNlrdRegModelXtraepochs', history.history['accuracy'][-1], history.history['val_accuracy'][-1], history.history['loss'][-1], history.history['val_loss'][-1], history]

Epoch 1/350

2023-11-26 15:03:47.938189: E tensorflow/core/grappler/optimizers/meta_optimizer.cc:954] layout failed: INVALID_ARGUMENT: Size of values 0 does not match size of permutation 4 @ fanin shape insequential_24/dropout_34/dropout/SelectV2-2-TransposeNHWCToNCHW-LayoutOptimizer

283/283 - 6s - loss: 3.0678 - accuracy: 0.1838 - val_loss: 2.7241 - val_accuracy: 0.1523 - lr: 0.0010 - 6s/epoch - 20ms/step
Epoch 2/350
283/283 - 3s - loss: 2.5944 - accuracy: 0.2735 - val_loss: 2.4719 - val_accuracy: 0.2387 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 3/350
283/283 - 3s - loss: 2.3924 - accuracy: 0.3473 - val_loss: 2.2890 - val_accuracy: 0.3460 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 4/350
283/283 - 3s - loss: 2.1321 - accuracy: 0.4159 - val_loss: 1.9372 - val_accuracy: 0.4767 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 5/350
283/283 - 3s - loss: 1.9845 - accuracy: 0.4696 - val_loss: 1.8172 - val_accuracy: 0.5180 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 6/350
283/283 - 3s - loss: 1.8690 - accuracy: 0.5209 - val_loss: 1.7352 - val_accuracy: 0.5500 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 7/350
283/283 - 3s - loss: 1.8245 - accuracy: 0.5394 - val_loss: 2.0247 - val_accuracy: 0.4900 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 8/350
283/283 - 3s - loss: 1.8607 - accuracy: 0.5418 - val_loss: 1.6863 - val_accuracy: 0.6077 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 9/350
283/283 - 3s - loss: 1.7279 - accuracy: 0.5885 - val_loss: 1.7269 - val_accuracy: 0.6060 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 10/350
283/283 - 3s - loss: 1.6987 - accuracy: 0.6076 - val_loss: 1.6248 - val_accuracy: 0.6460 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 11/350
283/283 - 3s - loss: 1.6723 - accuracy: 0.6171 - val_loss: 1.5888 - val_accuracy: 0.6673 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 12/350
283/283 - 3s - loss: 1.6135 - accuracy: 0.6399 - val_loss: 1.6892 - val_accuracy: 0.6270 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 13/350
283/283 - 3s - loss: 1.6639 - accuracy: 0.6346 - val_loss: 1.6537 - val_accuracy: 0.6653 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 14/350
283/283 - 3s - loss: 1.6045 - accuracy: 0.6616 - val_loss: 1.5393 - val_accuracy: 0.6980 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 15/350
283/283 - 3s - loss: 1.5672 - accuracy: 0.6787 - val_loss: 1.6009 - val_accuracy: 0.6883 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 16/350
283/283 - 3s - loss: 1.5699 - accuracy: 0.6787 - val_loss: 1.6534 - val_accuracy: 0.6770 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 17/350
283/283 - 3s - loss: 1.5803 - accuracy: 0.6824 - val_loss: 1.5131 - val_accuracy: 0.7163 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 18/350
283/283 - 3s - loss: 1.5708 - accuracy: 0.6895 - val_loss: 1.7273 - val_accuracy: 0.6590 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 19/350
283/283 - 3s - loss: 1.5618 - accuracy: 0.6989 - val_loss: 1.4860 - val_accuracy: 0.7473 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 20/350
283/283 - 3s - loss: 1.5139 - accuracy: 0.7103 - val_loss: 1.4590 - val_accuracy: 0.7443 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 21/350
283/283 - 3s - loss: 1.5051 - accuracy: 0.7180 - val_loss: 1.4885 - val_accuracy: 0.7363 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 22/350
283/283 - 3s - loss: 1.4846 - accuracy: 0.7183 - val_loss: 1.5989 - val_accuracy: 0.7043 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 23/350
283/283 - 3s - loss: 1.4912 - accuracy: 0.7247 - val_loss: 1.5380 - val_accuracy: 0.7340 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 24/350
283/283 - 3s - loss: 1.5027 - accuracy: 0.7219 - val_loss: 1.4953 - val_accuracy: 0.7413 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 25/350
283/283 - 3s - loss: 1.4732 - accuracy: 0.7362 - val_loss: 1.5018 - val_accuracy: 0.7457 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 26/350
283/283 - 3s - loss: 1.4799 - accuracy: 0.7306 - val_loss: 1.5084 - val_accuracy: 0.7450 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 27/350
283/283 - 3s - loss: 1.4329 - accuracy: 0.7476 - val_loss: 1.6370 - val_accuracy: 0.7243 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 28/350
283/283 - 3s - loss: 1.4888 - accuracy: 0.7438 - val_loss: 1.4910 - val_accuracy: 0.7507 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 29/350
283/283 - 3s - loss: 1.4593 - accuracy: 0.7510 - val_loss: 2.0608 - val_accuracy: 0.5950 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 30/350

Epoch 30: ReduceLROnPlateau reducing learning rate to 0.000800000037997961.
283/283 - 3s - loss: 1.4488 - accuracy: 0.7525 - val_loss: 1.5298 - val_accuracy: 0.7507 - lr: 0.0010 - 3s/epoch - 9ms/step
Epoch 31/350
283/283 - 3s - loss: 1.4048 - accuracy: 0.7630 - val_loss: 1.3696 - val_accuracy: 0.7837 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 32/350
283/283 - 3s - loss: 1.3253 - accuracy: 0.7759 - val_loss: 1.3595 - val_accuracy: 0.7827 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 33/350
283/283 - 3s - loss: 1.2966 - accuracy: 0.7794 - val_loss: 1.3295 - val_accuracy: 0.7867 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 34/350
283/283 - 3s - loss: 1.2828 - accuracy: 0.7848 - val_loss: 1.4173 - val_accuracy: 0.7597 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 35/350
283/283 - 3s - loss: 1.2698 - accuracy: 0.7893 - val_loss: 1.3647 - val_accuracy: 0.7713 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 36/350
283/283 - 3s - loss: 1.2532 - accuracy: 0.7871 - val_loss: 1.3976 - val_accuracy: 0.7627 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 37/350
283/283 - 3s - loss: 1.2622 - accuracy: 0.7839 - val_loss: 1.3490 - val_accuracy: 0.7710 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 38/350
283/283 - 3s - loss: 1.2640 - accuracy: 0.7840 - val_loss: 1.3227 - val_accuracy: 0.7797 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 39/350
283/283 - 3s - loss: 1.2482 - accuracy: 0.7914 - val_loss: 1.3562 - val_accuracy: 0.7767 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 40/350
283/283 - 3s - loss: 1.2499 - accuracy: 0.7900 - val_loss: 1.3259 - val_accuracy: 0.7813 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 41/350
283/283 - 2s - loss: 1.2324 - accuracy: 0.7992 - val_loss: 1.3595 - val_accuracy: 0.7727 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 42/350
283/283 - 3s - loss: 1.2158 - accuracy: 0.8036 - val_loss: 1.2406 - val_accuracy: 0.8047 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 43/350
283/283 - 3s - loss: 1.2034 - accuracy: 0.8021 - val_loss: 1.2210 - val_accuracy: 0.8123 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 44/350
283/283 - 3s - loss: 1.2341 - accuracy: 0.7966 - val_loss: 1.2540 - val_accuracy: 0.8063 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 45/350
283/283 - 3s - loss: 1.1996 - accuracy: 0.8139 - val_loss: 1.3065 - val_accuracy: 0.7897 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 46/350
283/283 - 2s - loss: 1.1893 - accuracy: 0.8123 - val_loss: 1.2650 - val_accuracy: 0.8007 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 47/350
283/283 - 2s - loss: 1.2023 - accuracy: 0.8104 - val_loss: 1.3176 - val_accuracy: 0.7957 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 48/350
283/283 - 3s - loss: 1.2029 - accuracy: 0.8089 - val_loss: 1.2689 - val_accuracy: 0.8100 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 49/350
283/283 - 2s - loss: 1.1976 - accuracy: 0.8095 - val_loss: 1.3170 - val_accuracy: 0.7817 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 50/350
283/283 - 3s - loss: 1.1751 - accuracy: 0.8191 - val_loss: 1.2102 - val_accuracy: 0.8217 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 51/350
283/283 - 3s - loss: 1.1532 - accuracy: 0.8234 - val_loss: 1.2187 - val_accuracy: 0.8170 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 52/350
283/283 - 3s - loss: 1.1291 - accuracy: 0.8263 - val_loss: 1.2663 - val_accuracy: 0.7987 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 53/350
283/283 - 3s - loss: 1.1368 - accuracy: 0.8296 - val_loss: 1.2303 - val_accuracy: 0.8150 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 54/350
283/283 - 3s - loss: 1.1897 - accuracy: 0.8178 - val_loss: 1.2587 - val_accuracy: 0.8070 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 55/350
283/283 - 3s - loss: 1.1996 - accuracy: 0.8185 - val_loss: 1.2255 - val_accuracy: 0.8257 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 56/350
283/283 - 3s - loss: 1.1518 - accuracy: 0.8268 - val_loss: 1.2198 - val_accuracy: 0.8177 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 57/350
283/283 - 3s - loss: 1.1655 - accuracy: 0.8245 - val_loss: 1.2061 - val_accuracy: 0.8207 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 58/350
283/283 - 3s - loss: 1.1707 - accuracy: 0.8237 - val_loss: 1.2282 - val_accuracy: 0.8160 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 59/350
283/283 - 3s - loss: 1.1113 - accuracy: 0.8356 - val_loss: 1.2952 - val_accuracy: 0.7940 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 60/350
283/283 - 3s - loss: 1.1404 - accuracy: 0.8302 - val_loss: 1.2658 - val_accuracy: 0.7913 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 61/350
283/283 - 3s - loss: 1.1408 - accuracy: 0.8365 - val_loss: 1.1905 - val_accuracy: 0.8287 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 62/350
283/283 - 3s - loss: 1.1338 - accuracy: 0.8310 - val_loss: 1.2151 - val_accuracy: 0.8190 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 63/350
283/283 - 3s - loss: 1.1307 - accuracy: 0.8350 - val_loss: 1.2305 - val_accuracy: 0.8183 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 64/350
283/283 - 3s - loss: 1.1278 - accuracy: 0.8370 - val_loss: 1.1761 - val_accuracy: 0.8307 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 65/350
283/283 - 3s - loss: 1.1164 - accuracy: 0.8409 - val_loss: 1.2995 - val_accuracy: 0.7937 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 66/350
283/283 - 3s - loss: 1.1200 - accuracy: 0.8405 - val_loss: 1.2103 - val_accuracy: 0.8233 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 67/350
283/283 - 3s - loss: 1.1806 - accuracy: 0.8271 - val_loss: 1.2946 - val_accuracy: 0.7997 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 68/350
283/283 - 3s - loss: 1.1100 - accuracy: 0.8477 - val_loss: 1.3410 - val_accuracy: 0.7937 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 69/350
283/283 - 3s - loss: 1.1019 - accuracy: 0.8480 - val_loss: 1.2960 - val_accuracy: 0.8000 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 70/350
283/283 - 3s - loss: 1.0846 - accuracy: 0.8480 - val_loss: 1.2126 - val_accuracy: 0.8187 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 71/350
283/283 - 3s - loss: 1.1277 - accuracy: 0.8395 - val_loss: 1.3035 - val_accuracy: 0.7990 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 72/350
283/283 - 3s - loss: 1.0943 - accuracy: 0.8463 - val_loss: 1.2179 - val_accuracy: 0.8210 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 73/350
283/283 - 3s - loss: 1.0950 - accuracy: 0.8492 - val_loss: 1.1453 - val_accuracy: 0.8440 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 74/350
283/283 - 3s - loss: 1.0919 - accuracy: 0.8481 - val_loss: 1.1126 - val_accuracy: 0.8507 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 75/350
283/283 - 3s - loss: 1.0863 - accuracy: 0.8530 - val_loss: 1.1431 - val_accuracy: 0.8380 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 76/350
283/283 - 3s - loss: 1.0682 - accuracy: 0.8595 - val_loss: 1.1599 - val_accuracy: 0.8353 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 77/350
283/283 - 3s - loss: 1.0661 - accuracy: 0.8573 - val_loss: 1.3301 - val_accuracy: 0.7817 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 78/350
283/283 - 3s - loss: 1.0854 - accuracy: 0.8479 - val_loss: 1.1736 - val_accuracy: 0.8313 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 79/350
283/283 - 3s - loss: 1.0747 - accuracy: 0.8531 - val_loss: 1.1313 - val_accuracy: 0.8463 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 80/350
283/283 - 2s - loss: 1.0660 - accuracy: 0.8598 - val_loss: 1.1325 - val_accuracy: 0.8397 - lr: 8.0000e-04 - 2s/epoch - 9ms/step
Epoch 81/350
283/283 - 3s - loss: 1.0706 - accuracy: 0.8539 - val_loss: 1.3071 - val_accuracy: 0.7973 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 82/350
283/283 - 3s - loss: 1.1058 - accuracy: 0.8515 - val_loss: 1.1486 - val_accuracy: 0.8417 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 83/350
283/283 - 3s - loss: 1.0702 - accuracy: 0.8618 - val_loss: 1.1703 - val_accuracy: 0.8363 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 84/350

Epoch 84: ReduceLROnPlateau reducing learning rate to 0.0006400000303983689.
283/283 - 3s - loss: 1.0828 - accuracy: 0.8530 - val_loss: 1.1348 - val_accuracy: 0.8430 - lr: 8.0000e-04 - 3s/epoch - 9ms/step
Epoch 85/350
283/283 - 3s - loss: 1.0097 - accuracy: 0.8743 - val_loss: 1.1019 - val_accuracy: 0.8520 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 86/350
283/283 - 3s - loss: 0.9840 - accuracy: 0.8692 - val_loss: 1.1048 - val_accuracy: 0.8363 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 87/350
283/283 - 3s - loss: 0.9603 - accuracy: 0.8761 - val_loss: 1.0191 - val_accuracy: 0.8643 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 88/350
283/283 - 3s - loss: 0.9285 - accuracy: 0.8796 - val_loss: 1.0421 - val_accuracy: 0.8520 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 89/350
283/283 - 3s - loss: 0.9061 - accuracy: 0.8789 - val_loss: 1.0636 - val_accuracy: 0.8423 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 90/350
283/283 - 3s - loss: 0.8900 - accuracy: 0.8844 - val_loss: 0.9970 - val_accuracy: 0.8543 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 91/350
283/283 - 3s - loss: 0.8951 - accuracy: 0.8850 - val_loss: 1.0180 - val_accuracy: 0.8507 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 92/350
283/283 - 3s - loss: 0.8991 - accuracy: 0.8798 - val_loss: 1.0900 - val_accuracy: 0.8333 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 93/350
283/283 - 3s - loss: 0.8920 - accuracy: 0.8826 - val_loss: 0.9770 - val_accuracy: 0.8620 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 94/350
283/283 - 3s - loss: 0.9159 - accuracy: 0.8696 - val_loss: 1.0327 - val_accuracy: 0.8363 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 95/350
283/283 - 3s - loss: 0.9314 - accuracy: 0.8682 - val_loss: 1.0597 - val_accuracy: 0.8443 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 96/350
283/283 - 3s - loss: 0.9039 - accuracy: 0.8809 - val_loss: 1.0084 - val_accuracy: 0.8513 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 97/350
283/283 - 3s - loss: 0.8945 - accuracy: 0.8800 - val_loss: 0.9613 - val_accuracy: 0.8740 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 98/350
283/283 - 3s - loss: 0.8927 - accuracy: 0.8836 - val_loss: 0.9751 - val_accuracy: 0.8603 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 99/350
283/283 - 3s - loss: 0.8744 - accuracy: 0.8877 - val_loss: 1.1536 - val_accuracy: 0.8160 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 100/350
283/283 - 3s - loss: 0.9015 - accuracy: 0.8747 - val_loss: 0.9730 - val_accuracy: 0.8613 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 101/350
283/283 - 3s - loss: 0.8719 - accuracy: 0.8855 - val_loss: 1.0586 - val_accuracy: 0.8400 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 102/350
283/283 - 3s - loss: 0.8829 - accuracy: 0.8788 - val_loss: 0.9849 - val_accuracy: 0.8580 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 103/350
283/283 - 3s - loss: 0.8495 - accuracy: 0.8921 - val_loss: 1.0131 - val_accuracy: 0.8423 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 104/350
283/283 - 3s - loss: 0.8616 - accuracy: 0.8874 - val_loss: 1.0055 - val_accuracy: 0.8500 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 105/350
283/283 - 3s - loss: 0.9085 - accuracy: 0.8776 - val_loss: 1.0996 - val_accuracy: 0.8270 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 106/350
283/283 - 3s - loss: 0.8719 - accuracy: 0.8826 - val_loss: 0.9756 - val_accuracy: 0.8620 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 107/350

Epoch 107: ReduceLROnPlateau reducing learning rate to 0.0005120000336319208.
283/283 - 3s - loss: 0.8718 - accuracy: 0.8827 - val_loss: 0.9651 - val_accuracy: 0.8653 - lr: 6.4000e-04 - 3s/epoch - 9ms/step
Epoch 108/350
283/283 - 3s - loss: 0.8321 - accuracy: 0.8934 - val_loss: 0.9585 - val_accuracy: 0.8647 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 109/350
283/283 - 3s - loss: 0.7983 - accuracy: 0.8983 - val_loss: 0.9047 - val_accuracy: 0.8733 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 110/350
283/283 - 3s - loss: 0.7932 - accuracy: 0.8951 - val_loss: 0.9143 - val_accuracy: 0.8710 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 111/350
283/283 - 3s - loss: 0.7969 - accuracy: 0.8938 - val_loss: 0.9441 - val_accuracy: 0.8617 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 112/350
283/283 - 3s - loss: 0.7698 - accuracy: 0.9013 - val_loss: 0.9237 - val_accuracy: 0.8700 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 113/350
283/283 - 3s - loss: 0.7506 - accuracy: 0.9033 - val_loss: 0.8917 - val_accuracy: 0.8690 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 114/350
283/283 - 3s - loss: 0.7497 - accuracy: 0.9046 - val_loss: 0.9437 - val_accuracy: 0.8517 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 115/350
283/283 - 3s - loss: 0.7892 - accuracy: 0.8928 - val_loss: 0.8977 - val_accuracy: 0.8657 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 116/350
283/283 - 3s - loss: 0.7576 - accuracy: 0.8982 - val_loss: 0.8838 - val_accuracy: 0.8657 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 117/350
283/283 - 3s - loss: 0.7530 - accuracy: 0.9014 - val_loss: 0.8943 - val_accuracy: 0.8633 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 118/350
283/283 - 3s - loss: 0.7471 - accuracy: 0.8984 - val_loss: 0.8678 - val_accuracy: 0.8710 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 119/350
283/283 - 3s - loss: 0.7473 - accuracy: 0.8983 - val_loss: 0.9464 - val_accuracy: 0.8433 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 120/350
283/283 - 3s - loss: 0.7271 - accuracy: 0.9036 - val_loss: 0.8913 - val_accuracy: 0.8653 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 121/350
283/283 - 3s - loss: 0.7370 - accuracy: 0.9033 - val_loss: 0.9273 - val_accuracy: 0.8573 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 122/350
283/283 - 3s - loss: 0.7202 - accuracy: 0.9054 - val_loss: 0.8282 - val_accuracy: 0.8797 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 123/350
283/283 - 3s - loss: 0.7204 - accuracy: 0.9043 - val_loss: 0.9015 - val_accuracy: 0.8633 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 124/350
283/283 - 3s - loss: 0.7086 - accuracy: 0.9083 - val_loss: 0.8771 - val_accuracy: 0.8640 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 125/350
283/283 - 3s - loss: 0.7271 - accuracy: 0.9062 - val_loss: 0.8563 - val_accuracy: 0.8633 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 126/350
283/283 - 3s - loss: 0.7287 - accuracy: 0.9006 - val_loss: 0.8643 - val_accuracy: 0.8627 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 127/350
283/283 - 3s - loss: 0.7441 - accuracy: 0.8943 - val_loss: 0.8700 - val_accuracy: 0.8660 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 128/350
283/283 - 3s - loss: 0.7240 - accuracy: 0.9040 - val_loss: 0.9665 - val_accuracy: 0.8373 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 129/350
283/283 - 3s - loss: 0.7223 - accuracy: 0.9055 - val_loss: 0.8475 - val_accuracy: 0.8723 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 130/350
283/283 - 3s - loss: 0.6810 - accuracy: 0.9098 - val_loss: 0.8566 - val_accuracy: 0.8703 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 131/350
283/283 - 3s - loss: 0.6938 - accuracy: 0.9109 - val_loss: 0.8140 - val_accuracy: 0.8823 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 132/350
283/283 - 3s - loss: 0.7174 - accuracy: 0.9055 - val_loss: 0.8740 - val_accuracy: 0.8637 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 133/350
283/283 - 3s - loss: 0.7122 - accuracy: 0.9071 - val_loss: 0.8318 - val_accuracy: 0.8750 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 134/350
283/283 - 3s - loss: 0.6780 - accuracy: 0.9125 - val_loss: 0.8813 - val_accuracy: 0.8570 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 135/350
283/283 - 3s - loss: 0.6840 - accuracy: 0.9112 - val_loss: 0.9079 - val_accuracy: 0.8523 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 136/350
283/283 - 3s - loss: 0.6913 - accuracy: 0.9086 - val_loss: 0.8699 - val_accuracy: 0.8600 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 137/350
283/283 - 3s - loss: 0.7171 - accuracy: 0.9020 - val_loss: 0.8355 - val_accuracy: 0.8650 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 138/350
283/283 - 3s - loss: 0.7079 - accuracy: 0.9035 - val_loss: 0.7977 - val_accuracy: 0.8853 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 139/350
283/283 - 3s - loss: 0.6867 - accuracy: 0.9127 - val_loss: 0.8167 - val_accuracy: 0.8837 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 140/350
283/283 - 3s - loss: 0.6878 - accuracy: 0.9102 - val_loss: 0.8267 - val_accuracy: 0.8757 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 141/350
283/283 - 3s - loss: 0.7101 - accuracy: 0.9095 - val_loss: 0.8036 - val_accuracy: 0.8770 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 142/350
283/283 - 3s - loss: 0.7072 - accuracy: 0.9070 - val_loss: 0.9101 - val_accuracy: 0.8563 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 143/350
283/283 - 3s - loss: 0.6939 - accuracy: 0.9091 - val_loss: 0.8549 - val_accuracy: 0.8680 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 144/350
283/283 - 2s - loss: 0.7000 - accuracy: 0.9084 - val_loss: 0.9772 - val_accuracy: 0.8300 - lr: 5.1200e-04 - 2s/epoch - 9ms/step
Epoch 145/350
283/283 - 3s - loss: 0.7061 - accuracy: 0.9037 - val_loss: 0.8522 - val_accuracy: 0.8737 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 146/350
283/283 - 3s - loss: 0.7293 - accuracy: 0.9045 - val_loss: 0.8383 - val_accuracy: 0.8740 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 147/350
283/283 - 3s - loss: 0.7122 - accuracy: 0.9051 - val_loss: 0.9163 - val_accuracy: 0.8580 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 148/350

Epoch 148: ReduceLROnPlateau reducing learning rate to 0.00040960004553198815.
283/283 - 3s - loss: 0.6876 - accuracy: 0.9140 - val_loss: 0.8071 - val_accuracy: 0.8843 - lr: 5.1200e-04 - 3s/epoch - 9ms/step
Epoch 149/350
283/283 - 3s - loss: 0.6698 - accuracy: 0.9154 - val_loss: 0.8027 - val_accuracy: 0.8807 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 150/350
283/283 - 3s - loss: 0.6589 - accuracy: 0.9155 - val_loss: 0.7703 - val_accuracy: 0.8863 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 151/350
283/283 - 3s - loss: 0.6361 - accuracy: 0.9195 - val_loss: 0.8041 - val_accuracy: 0.8770 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 152/350
283/283 - 3s - loss: 0.6233 - accuracy: 0.9204 - val_loss: 0.7623 - val_accuracy: 0.8910 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 153/350
283/283 - 3s - loss: 0.6094 - accuracy: 0.9225 - val_loss: 0.8044 - val_accuracy: 0.8760 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 154/350
283/283 - 3s - loss: 0.6495 - accuracy: 0.9113 - val_loss: 0.7921 - val_accuracy: 0.8753 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 155/350
283/283 - 3s - loss: 0.6054 - accuracy: 0.9265 - val_loss: 0.7792 - val_accuracy: 0.8787 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 156/350
283/283 - 3s - loss: 0.6167 - accuracy: 0.9143 - val_loss: 0.7737 - val_accuracy: 0.8790 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 157/350
283/283 - 3s - loss: 0.6063 - accuracy: 0.9204 - val_loss: 0.7744 - val_accuracy: 0.8763 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 158/350
283/283 - 3s - loss: 0.5903 - accuracy: 0.9270 - val_loss: 0.7407 - val_accuracy: 0.8837 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 159/350
283/283 - 3s - loss: 0.5957 - accuracy: 0.9193 - val_loss: 0.7575 - val_accuracy: 0.8793 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 160/350
283/283 - 3s - loss: 0.5804 - accuracy: 0.9280 - val_loss: 0.7542 - val_accuracy: 0.8767 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 161/350
283/283 - 3s - loss: 0.5854 - accuracy: 0.9216 - val_loss: 0.7340 - val_accuracy: 0.8827 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 162/350
283/283 - 3s - loss: 0.5938 - accuracy: 0.9195 - val_loss: 0.7934 - val_accuracy: 0.8780 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 163/350
283/283 - 3s - loss: 0.5779 - accuracy: 0.9240 - val_loss: 0.8082 - val_accuracy: 0.8677 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 164/350
283/283 - 3s - loss: 0.5972 - accuracy: 0.9189 - val_loss: 0.7622 - val_accuracy: 0.8770 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 165/350
283/283 - 3s - loss: 0.5853 - accuracy: 0.9233 - val_loss: 0.7414 - val_accuracy: 0.8853 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 166/350
283/283 - 3s - loss: 0.5843 - accuracy: 0.9214 - val_loss: 0.7512 - val_accuracy: 0.8797 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 167/350
283/283 - 3s - loss: 0.5697 - accuracy: 0.9251 - val_loss: 0.7182 - val_accuracy: 0.8910 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 168/350
283/283 - 3s - loss: 0.5950 - accuracy: 0.9194 - val_loss: 0.7869 - val_accuracy: 0.8647 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 169/350
283/283 - 3s - loss: 0.6084 - accuracy: 0.9139 - val_loss: 0.7630 - val_accuracy: 0.8773 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 170/350
283/283 - 3s - loss: 0.5937 - accuracy: 0.9204 - val_loss: 0.7731 - val_accuracy: 0.8737 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 171/350
283/283 - 3s - loss: 0.5754 - accuracy: 0.9243 - val_loss: 0.7545 - val_accuracy: 0.8827 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 172/350
283/283 - 3s - loss: 0.5723 - accuracy: 0.9238 - val_loss: 0.7308 - val_accuracy: 0.8897 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 173/350
283/283 - 3s - loss: 0.6062 - accuracy: 0.9168 - val_loss: 0.7219 - val_accuracy: 0.8907 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 174/350
283/283 - 3s - loss: 0.5687 - accuracy: 0.9248 - val_loss: 0.8957 - val_accuracy: 0.8360 - lr: 4.0960e-04 - 3s/epoch - 9ms/step
Epoch 175/350
283/283 - 2s - loss: 0.6160 - accuracy: 0.9136 - val_loss: 0.7217 - val_accuracy: 0.8917 - lr: 4.0960e-04 - 2s/epoch - 9ms/step
Epoch 176/350
283/283 - 2s - loss: 0.5596 - accuracy: 0.9290 - val_loss: 0.7364 - val_accuracy: 0.8887 - lr: 4.0960e-04 - 2s/epoch - 9ms/step
Epoch 177/350

Epoch 177: ReduceLROnPlateau reducing learning rate to 0.00032768002711236477.
283/283 - 2s - loss: 0.5754 - accuracy: 0.9277 - val_loss: 0.7204 - val_accuracy: 0.8887 - lr: 4.0960e-04 - 2s/epoch - 9ms/step
Epoch 178/350
283/283 - 3s - loss: 0.5450 - accuracy: 0.9300 - val_loss: 0.6882 - val_accuracy: 0.8943 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 179/350
283/283 - 3s - loss: 0.5647 - accuracy: 0.9240 - val_loss: 0.7039 - val_accuracy: 0.8867 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 180/350
283/283 - 3s - loss: 0.5575 - accuracy: 0.9246 - val_loss: 0.7175 - val_accuracy: 0.8880 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 181/350
283/283 - 3s - loss: 0.5256 - accuracy: 0.9352 - val_loss: 0.7409 - val_accuracy: 0.8803 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 182/350
283/283 - 3s - loss: 0.5444 - accuracy: 0.9284 - val_loss: 0.6910 - val_accuracy: 0.8897 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 183/350
283/283 - 3s - loss: 0.5179 - accuracy: 0.9343 - val_loss: 0.6978 - val_accuracy: 0.8880 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 184/350
283/283 - 3s - loss: 0.5156 - accuracy: 0.9314 - val_loss: 0.6843 - val_accuracy: 0.8893 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 185/350
283/283 - 3s - loss: 0.5029 - accuracy: 0.9332 - val_loss: 0.6881 - val_accuracy: 0.8870 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 186/350
283/283 - 3s - loss: 0.4932 - accuracy: 0.9376 - val_loss: 0.6669 - val_accuracy: 0.8913 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 187/350
283/283 - 3s - loss: 0.5032 - accuracy: 0.9324 - val_loss: 0.6979 - val_accuracy: 0.8860 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 188/350
283/283 - 3s - loss: 0.4757 - accuracy: 0.9406 - val_loss: 0.6950 - val_accuracy: 0.8790 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 189/350
283/283 - 3s - loss: 0.4925 - accuracy: 0.9354 - val_loss: 0.6724 - val_accuracy: 0.8910 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 190/350
283/283 - 3s - loss: 0.4990 - accuracy: 0.9343 - val_loss: 0.6659 - val_accuracy: 0.8923 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 191/350
283/283 - 3s - loss: 0.4900 - accuracy: 0.9360 - val_loss: 0.6814 - val_accuracy: 0.8830 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 192/350
283/283 - 3s - loss: 0.4827 - accuracy: 0.9343 - val_loss: 0.6601 - val_accuracy: 0.8953 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 193/350
283/283 - 3s - loss: 0.4795 - accuracy: 0.9374 - val_loss: 0.6938 - val_accuracy: 0.8820 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 194/350
283/283 - 3s - loss: 0.4987 - accuracy: 0.9325 - val_loss: 0.6491 - val_accuracy: 0.8920 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 195/350
283/283 - 3s - loss: 0.4795 - accuracy: 0.9360 - val_loss: 0.6781 - val_accuracy: 0.8817 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 196/350
283/283 - 3s - loss: 0.4802 - accuracy: 0.9327 - val_loss: 0.6893 - val_accuracy: 0.8797 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 197/350
283/283 - 3s - loss: 0.4775 - accuracy: 0.9358 - val_loss: 0.6701 - val_accuracy: 0.8877 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 198/350
283/283 - 3s - loss: 0.4881 - accuracy: 0.9321 - val_loss: 0.6972 - val_accuracy: 0.8820 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 199/350
283/283 - 3s - loss: 0.4878 - accuracy: 0.9332 - val_loss: 0.6827 - val_accuracy: 0.8843 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 200/350
283/283 - 3s - loss: 0.5079 - accuracy: 0.9302 - val_loss: 0.7860 - val_accuracy: 0.8500 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 201/350
283/283 - 3s - loss: 0.4943 - accuracy: 0.9344 - val_loss: 0.6590 - val_accuracy: 0.8963 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 202/350
283/283 - 3s - loss: 0.4771 - accuracy: 0.9380 - val_loss: 0.6472 - val_accuracy: 0.8863 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 203/350
283/283 - 3s - loss: 0.4726 - accuracy: 0.9383 - val_loss: 0.7513 - val_accuracy: 0.8670 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 204/350
283/283 - 3s - loss: 0.4753 - accuracy: 0.9363 - val_loss: 0.6650 - val_accuracy: 0.8890 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 205/350
283/283 - 3s - loss: 0.4663 - accuracy: 0.9381 - val_loss: 0.6848 - val_accuracy: 0.8820 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 206/350
283/283 - 3s - loss: 0.4836 - accuracy: 0.9358 - val_loss: 0.6256 - val_accuracy: 0.8937 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 207/350
283/283 - 2s - loss: 0.4637 - accuracy: 0.9377 - val_loss: 0.6880 - val_accuracy: 0.8790 - lr: 3.2768e-04 - 2s/epoch - 9ms/step
Epoch 208/350
283/283 - 2s - loss: 0.4775 - accuracy: 0.9356 - val_loss: 0.6673 - val_accuracy: 0.8900 - lr: 3.2768e-04 - 2s/epoch - 9ms/step
Epoch 209/350
283/283 - 3s - loss: 0.4537 - accuracy: 0.9410 - val_loss: 0.6393 - val_accuracy: 0.8947 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 210/350
283/283 - 3s - loss: 0.4693 - accuracy: 0.9391 - val_loss: 0.6549 - val_accuracy: 0.8927 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 211/350
283/283 - 3s - loss: 0.4721 - accuracy: 0.9391 - val_loss: 0.6385 - val_accuracy: 0.8900 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 212/350
283/283 - 3s - loss: 0.4799 - accuracy: 0.9337 - val_loss: 0.6682 - val_accuracy: 0.8893 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 213/350
283/283 - 3s - loss: 0.4826 - accuracy: 0.9330 - val_loss: 0.6395 - val_accuracy: 0.8893 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 214/350
283/283 - 3s - loss: 0.4598 - accuracy: 0.9435 - val_loss: 0.6470 - val_accuracy: 0.8897 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 215/350
283/283 - 3s - loss: 0.4752 - accuracy: 0.9351 - val_loss: 0.6350 - val_accuracy: 0.8960 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 216/350

Epoch 216: ReduceLROnPlateau reducing learning rate to 0.0002621440216898918.
283/283 - 3s - loss: 0.4818 - accuracy: 0.9328 - val_loss: 0.6456 - val_accuracy: 0.8953 - lr: 3.2768e-04 - 3s/epoch - 9ms/step
Epoch 217/350
283/283 - 3s - loss: 0.4669 - accuracy: 0.9390 - val_loss: 0.6547 - val_accuracy: 0.8883 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 218/350
283/283 - 3s - loss: 0.4499 - accuracy: 0.9405 - val_loss: 0.6292 - val_accuracy: 0.8957 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 219/350
283/283 - 3s - loss: 0.4359 - accuracy: 0.9440 - val_loss: 0.6646 - val_accuracy: 0.8873 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 220/350
283/283 - 3s - loss: 0.4302 - accuracy: 0.9428 - val_loss: 0.6571 - val_accuracy: 0.8797 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 221/350
283/283 - 3s - loss: 0.4456 - accuracy: 0.9386 - val_loss: 0.5998 - val_accuracy: 0.8987 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 222/350
283/283 - 3s - loss: 0.4408 - accuracy: 0.9392 - val_loss: 0.6188 - val_accuracy: 0.8920 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 223/350
283/283 - 3s - loss: 0.4275 - accuracy: 0.9440 - val_loss: 0.6119 - val_accuracy: 0.8990 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 224/350
283/283 - 3s - loss: 0.4150 - accuracy: 0.9445 - val_loss: 0.6681 - val_accuracy: 0.8817 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 225/350
283/283 - 3s - loss: 0.4248 - accuracy: 0.9405 - val_loss: 0.5992 - val_accuracy: 0.8983 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 226/350
283/283 - 3s - loss: 0.4124 - accuracy: 0.9458 - val_loss: 0.6519 - val_accuracy: 0.8810 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 227/350
283/283 - 3s - loss: 0.4113 - accuracy: 0.9427 - val_loss: 0.6308 - val_accuracy: 0.8957 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 228/350
283/283 - 3s - loss: 0.4151 - accuracy: 0.9437 - val_loss: 0.6207 - val_accuracy: 0.8890 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 229/350
283/283 - 3s - loss: 0.4233 - accuracy: 0.9423 - val_loss: 0.5957 - val_accuracy: 0.9010 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 230/350
283/283 - 3s - loss: 0.4002 - accuracy: 0.9443 - val_loss: 0.6133 - val_accuracy: 0.8943 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 231/350
283/283 - 3s - loss: 0.4073 - accuracy: 0.9423 - val_loss: 0.6044 - val_accuracy: 0.8953 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 232/350
283/283 - 3s - loss: 0.4098 - accuracy: 0.9417 - val_loss: 0.5869 - val_accuracy: 0.9040 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 233/350
283/283 - 3s - loss: 0.4198 - accuracy: 0.9406 - val_loss: 0.6292 - val_accuracy: 0.8873 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 234/350
283/283 - 3s - loss: 0.4073 - accuracy: 0.9451 - val_loss: 0.6347 - val_accuracy: 0.8810 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 235/350
283/283 - 3s - loss: 0.4118 - accuracy: 0.9402 - val_loss: 0.5860 - val_accuracy: 0.8973 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 236/350
283/283 - 3s - loss: 0.3988 - accuracy: 0.9468 - val_loss: 0.6072 - val_accuracy: 0.8963 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 237/350
283/283 - 3s - loss: 0.3920 - accuracy: 0.9476 - val_loss: 0.6060 - val_accuracy: 0.8917 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 238/350
283/283 - 3s - loss: 0.4029 - accuracy: 0.9423 - val_loss: 0.5898 - val_accuracy: 0.8993 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 239/350
283/283 - 2s - loss: 0.3998 - accuracy: 0.9414 - val_loss: 0.6187 - val_accuracy: 0.8923 - lr: 2.6214e-04 - 2s/epoch - 9ms/step
Epoch 240/350
283/283 - 2s - loss: 0.3920 - accuracy: 0.9473 - val_loss: 0.6079 - val_accuracy: 0.9000 - lr: 2.6214e-04 - 2s/epoch - 9ms/step
Epoch 241/350
283/283 - 3s - loss: 0.4030 - accuracy: 0.9421 - val_loss: 0.5726 - val_accuracy: 0.9047 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 242/350
283/283 - 3s - loss: 0.3903 - accuracy: 0.9457 - val_loss: 0.5888 - val_accuracy: 0.8963 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 243/350
283/283 - 3s - loss: 0.3923 - accuracy: 0.9457 - val_loss: 0.5999 - val_accuracy: 0.8953 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 244/350
283/283 - 3s - loss: 0.3877 - accuracy: 0.9479 - val_loss: 0.5885 - val_accuracy: 0.8907 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 245/350
283/283 - 3s - loss: 0.3991 - accuracy: 0.9420 - val_loss: 0.6032 - val_accuracy: 0.8920 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 246/350
283/283 - 3s - loss: 0.3940 - accuracy: 0.9451 - val_loss: 0.5862 - val_accuracy: 0.9000 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 247/350
283/283 - 3s - loss: 0.3843 - accuracy: 0.9474 - val_loss: 0.5933 - val_accuracy: 0.8960 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 248/350
283/283 - 3s - loss: 0.4161 - accuracy: 0.9411 - val_loss: 0.5760 - val_accuracy: 0.8967 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 249/350
283/283 - 3s - loss: 0.3826 - accuracy: 0.9462 - val_loss: 0.5884 - val_accuracy: 0.8987 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 250/350
283/283 - 3s - loss: 0.4022 - accuracy: 0.9428 - val_loss: 0.5911 - val_accuracy: 0.8993 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 251/350

Epoch 251: ReduceLROnPlateau reducing learning rate to 0.00020971521735191345.
283/283 - 3s - loss: 0.3931 - accuracy: 0.9457 - val_loss: 0.6014 - val_accuracy: 0.8913 - lr: 2.6214e-04 - 3s/epoch - 9ms/step
Epoch 252/350
283/283 - 3s - loss: 0.3897 - accuracy: 0.9463 - val_loss: 0.5915 - val_accuracy: 0.8993 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 253/350
283/283 - 3s - loss: 0.3750 - accuracy: 0.9499 - val_loss: 0.5947 - val_accuracy: 0.8980 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 254/350
283/283 - 3s - loss: 0.3813 - accuracy: 0.9478 - val_loss: 0.5825 - val_accuracy: 0.8957 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 255/350
283/283 - 3s - loss: 0.3704 - accuracy: 0.9509 - val_loss: 0.6035 - val_accuracy: 0.8877 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 256/350
283/283 - 3s - loss: 0.3783 - accuracy: 0.9498 - val_loss: 0.5776 - val_accuracy: 0.8957 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 257/350
283/283 - 3s - loss: 0.3610 - accuracy: 0.9526 - val_loss: 0.5764 - val_accuracy: 0.8963 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 258/350
283/283 - 3s - loss: 0.3569 - accuracy: 0.9493 - val_loss: 0.5464 - val_accuracy: 0.9013 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 259/350
283/283 - 3s - loss: 0.3630 - accuracy: 0.9496 - val_loss: 0.5610 - val_accuracy: 0.8983 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 260/350
283/283 - 3s - loss: 0.3538 - accuracy: 0.9508 - val_loss: 0.5729 - val_accuracy: 0.8977 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 261/350
283/283 - 3s - loss: 0.3617 - accuracy: 0.9477 - val_loss: 0.5907 - val_accuracy: 0.8873 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 262/350
283/283 - 3s - loss: 0.3462 - accuracy: 0.9512 - val_loss: 0.5355 - val_accuracy: 0.9037 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 263/350
283/283 - 3s - loss: 0.3443 - accuracy: 0.9498 - val_loss: 0.5990 - val_accuracy: 0.8883 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 264/350
283/283 - 3s - loss: 0.3493 - accuracy: 0.9514 - val_loss: 0.5479 - val_accuracy: 0.9017 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 265/350
283/283 - 3s - loss: 0.3513 - accuracy: 0.9494 - val_loss: 0.5350 - val_accuracy: 0.9063 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 266/350
283/283 - 3s - loss: 0.3392 - accuracy: 0.9536 - val_loss: 0.5752 - val_accuracy: 0.8857 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 267/350
283/283 - 3s - loss: 0.3415 - accuracy: 0.9515 - val_loss: 0.5389 - val_accuracy: 0.9020 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 268/350
283/283 - 3s - loss: 0.3505 - accuracy: 0.9492 - val_loss: 0.5789 - val_accuracy: 0.8880 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 269/350
283/283 - 3s - loss: 0.3440 - accuracy: 0.9523 - val_loss: 0.5897 - val_accuracy: 0.8880 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 270/350
283/283 - 2s - loss: 0.3383 - accuracy: 0.9498 - val_loss: 0.5244 - val_accuracy: 0.9047 - lr: 2.0972e-04 - 2s/epoch - 9ms/step
Epoch 271/350
283/283 - 3s - loss: 0.3148 - accuracy: 0.9577 - val_loss: 0.5496 - val_accuracy: 0.8977 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 272/350
283/283 - 2s - loss: 0.3465 - accuracy: 0.9527 - val_loss: 0.5509 - val_accuracy: 0.9013 - lr: 2.0972e-04 - 2s/epoch - 9ms/step
Epoch 273/350
283/283 - 3s - loss: 0.3355 - accuracy: 0.9528 - val_loss: 0.5721 - val_accuracy: 0.8910 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 274/350
283/283 - 3s - loss: 0.3376 - accuracy: 0.9488 - val_loss: 0.5510 - val_accuracy: 0.8963 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 275/350
283/283 - 3s - loss: 0.3282 - accuracy: 0.9545 - val_loss: 0.5325 - val_accuracy: 0.9043 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 276/350
283/283 - 3s - loss: 0.3412 - accuracy: 0.9498 - val_loss: 0.5607 - val_accuracy: 0.8993 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 277/350
283/283 - 3s - loss: 0.3245 - accuracy: 0.9554 - val_loss: 0.5634 - val_accuracy: 0.8917 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 278/350
283/283 - 3s - loss: 0.3146 - accuracy: 0.9609 - val_loss: 0.5507 - val_accuracy: 0.9003 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 279/350
283/283 - 3s - loss: 0.3269 - accuracy: 0.9538 - val_loss: 0.5765 - val_accuracy: 0.8913 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 280/350

Epoch 280: ReduceLROnPlateau reducing learning rate to 0.00016777217388153076.
283/283 - 3s - loss: 0.3271 - accuracy: 0.9540 - val_loss: 0.5475 - val_accuracy: 0.8957 - lr: 2.0972e-04 - 3s/epoch - 9ms/step
Epoch 281/350
283/283 - 3s - loss: 0.3410 - accuracy: 0.9503 - val_loss: 0.5362 - val_accuracy: 0.9000 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 282/350
283/283 - 2s - loss: 0.3208 - accuracy: 0.9536 - val_loss: 0.5267 - val_accuracy: 0.8993 - lr: 1.6777e-04 - 2s/epoch - 9ms/step
Epoch 283/350
283/283 - 3s - loss: 0.3067 - accuracy: 0.9554 - val_loss: 0.5290 - val_accuracy: 0.9027 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 284/350
283/283 - 3s - loss: 0.3189 - accuracy: 0.9570 - val_loss: 0.5430 - val_accuracy: 0.8943 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 285/350
283/283 - 3s - loss: 0.3083 - accuracy: 0.9564 - val_loss: 0.5382 - val_accuracy: 0.8913 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 286/350
283/283 - 3s - loss: 0.3119 - accuracy: 0.9561 - val_loss: 0.5208 - val_accuracy: 0.9017 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 287/350
283/283 - 3s - loss: 0.3093 - accuracy: 0.9581 - val_loss: 0.6029 - val_accuracy: 0.8780 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 288/350
283/283 - 3s - loss: 0.3087 - accuracy: 0.9551 - val_loss: 0.4989 - val_accuracy: 0.9037 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 289/350
283/283 - 3s - loss: 0.3133 - accuracy: 0.9559 - val_loss: 0.4988 - val_accuracy: 0.9040 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 290/350
283/283 - 3s - loss: 0.2918 - accuracy: 0.9582 - val_loss: 0.4981 - val_accuracy: 0.9050 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 291/350
283/283 - 3s - loss: 0.2870 - accuracy: 0.9607 - val_loss: 0.4972 - val_accuracy: 0.9033 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 292/350
283/283 - 3s - loss: 0.2971 - accuracy: 0.9560 - val_loss: 0.5034 - val_accuracy: 0.9017 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 293/350
283/283 - 3s - loss: 0.3019 - accuracy: 0.9534 - val_loss: 0.5107 - val_accuracy: 0.8987 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 294/350
283/283 - 3s - loss: 0.2901 - accuracy: 0.9590 - val_loss: 0.5758 - val_accuracy: 0.8810 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 295/350
283/283 - 3s - loss: 0.3010 - accuracy: 0.9559 - val_loss: 0.5007 - val_accuracy: 0.9047 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 296/350
283/283 - 3s - loss: 0.2844 - accuracy: 0.9585 - val_loss: 0.5204 - val_accuracy: 0.8940 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 297/350
283/283 - 3s - loss: 0.2866 - accuracy: 0.9576 - val_loss: 0.5209 - val_accuracy: 0.9017 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 298/350
283/283 - 3s - loss: 0.2855 - accuracy: 0.9598 - val_loss: 0.5152 - val_accuracy: 0.9017 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 299/350
283/283 - 3s - loss: 0.2977 - accuracy: 0.9565 - val_loss: 0.4953 - val_accuracy: 0.9030 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 300/350
283/283 - 3s - loss: 0.2865 - accuracy: 0.9577 - val_loss: 0.5179 - val_accuracy: 0.8970 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 301/350
283/283 - 3s - loss: 0.2862 - accuracy: 0.9554 - val_loss: 0.5032 - val_accuracy: 0.8990 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 302/350
283/283 - 2s - loss: 0.2842 - accuracy: 0.9587 - val_loss: 0.5127 - val_accuracy: 0.9007 - lr: 1.6777e-04 - 2s/epoch - 9ms/step
Epoch 303/350
283/283 - 3s - loss: 0.2861 - accuracy: 0.9557 - val_loss: 0.5044 - val_accuracy: 0.8997 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 304/350
283/283 - 3s - loss: 0.2682 - accuracy: 0.9612 - val_loss: 0.5321 - val_accuracy: 0.8983 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 305/350
283/283 - 3s - loss: 0.2805 - accuracy: 0.9576 - val_loss: 0.4801 - val_accuracy: 0.9067 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 306/350
283/283 - 3s - loss: 0.2849 - accuracy: 0.9560 - val_loss: 0.5031 - val_accuracy: 0.9047 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 307/350
283/283 - 3s - loss: 0.2803 - accuracy: 0.9600 - val_loss: 0.5055 - val_accuracy: 0.9007 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 308/350
283/283 - 3s - loss: 0.2768 - accuracy: 0.9590 - val_loss: 0.5100 - val_accuracy: 0.9023 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 309/350
283/283 - 3s - loss: 0.2770 - accuracy: 0.9571 - val_loss: 0.4886 - val_accuracy: 0.9080 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 310/350
283/283 - 3s - loss: 0.2775 - accuracy: 0.9596 - val_loss: 0.5140 - val_accuracy: 0.9013 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 311/350
283/283 - 3s - loss: 0.2850 - accuracy: 0.9566 - val_loss: 0.5387 - val_accuracy: 0.8933 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 312/350
283/283 - 3s - loss: 0.2865 - accuracy: 0.9558 - val_loss: 0.5325 - val_accuracy: 0.8930 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 313/350
283/283 - 3s - loss: 0.2763 - accuracy: 0.9557 - val_loss: 0.5317 - val_accuracy: 0.8980 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 314/350
283/283 - 3s - loss: 0.2787 - accuracy: 0.9595 - val_loss: 0.4918 - val_accuracy: 0.9050 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 315/350

Epoch 315: ReduceLROnPlateau reducing learning rate to 0.00013421773910522462.
283/283 - 3s - loss: 0.2802 - accuracy: 0.9591 - val_loss: 0.4918 - val_accuracy: 0.9033 - lr: 1.6777e-04 - 3s/epoch - 9ms/step
Epoch 316/350
283/283 - 3s - loss: 0.2702 - accuracy: 0.9578 - val_loss: 0.4918 - val_accuracy: 0.9043 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 317/350
283/283 - 3s - loss: 0.2677 - accuracy: 0.9622 - val_loss: 0.5025 - val_accuracy: 0.9007 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 318/350
283/283 - 3s - loss: 0.2692 - accuracy: 0.9611 - val_loss: 0.4972 - val_accuracy: 0.9023 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 319/350
283/283 - 3s - loss: 0.2749 - accuracy: 0.9592 - val_loss: 0.4916 - val_accuracy: 0.9033 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 320/350
283/283 - 3s - loss: 0.2601 - accuracy: 0.9626 - val_loss: 0.4977 - val_accuracy: 0.9017 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 321/350
283/283 - 3s - loss: 0.2663 - accuracy: 0.9619 - val_loss: 0.4845 - val_accuracy: 0.9020 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 322/350
283/283 - 3s - loss: 0.2581 - accuracy: 0.9623 - val_loss: 0.4871 - val_accuracy: 0.9060 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 323/350
283/283 - 3s - loss: 0.2642 - accuracy: 0.9612 - val_loss: 0.4853 - val_accuracy: 0.9033 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 324/350
283/283 - 3s - loss: 0.2664 - accuracy: 0.9595 - val_loss: 0.4758 - val_accuracy: 0.9087 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 325/350
283/283 - 3s - loss: 0.2598 - accuracy: 0.9637 - val_loss: 0.5099 - val_accuracy: 0.9030 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 326/350
283/283 - 3s - loss: 0.2599 - accuracy: 0.9619 - val_loss: 0.5179 - val_accuracy: 0.8937 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 327/350
283/283 - 3s - loss: 0.2577 - accuracy: 0.9601 - val_loss: 0.5056 - val_accuracy: 0.8973 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 328/350
283/283 - 3s - loss: 0.2667 - accuracy: 0.9597 - val_loss: 0.4661 - val_accuracy: 0.9033 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 329/350
283/283 - 3s - loss: 0.2403 - accuracy: 0.9661 - val_loss: 0.4731 - val_accuracy: 0.9073 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 330/350
283/283 - 3s - loss: 0.2463 - accuracy: 0.9639 - val_loss: 0.4709 - val_accuracy: 0.9033 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 331/350
283/283 - 3s - loss: 0.2418 - accuracy: 0.9650 - val_loss: 0.4689 - val_accuracy: 0.9063 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 332/350
283/283 - 3s - loss: 0.2447 - accuracy: 0.9637 - val_loss: 0.4874 - val_accuracy: 0.9030 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 333/350
283/283 - 3s - loss: 0.2515 - accuracy: 0.9608 - val_loss: 0.4808 - val_accuracy: 0.9030 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 334/350
283/283 - 3s - loss: 0.2459 - accuracy: 0.9636 - val_loss: 0.5067 - val_accuracy: 0.8947 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 335/350
283/283 - 3s - loss: 0.2440 - accuracy: 0.9652 - val_loss: 0.4701 - val_accuracy: 0.9033 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 336/350
283/283 - 2s - loss: 0.2515 - accuracy: 0.9606 - val_loss: 0.4591 - val_accuracy: 0.9080 - lr: 1.3422e-04 - 2s/epoch - 9ms/step
Epoch 337/350
283/283 - 3s - loss: 0.2553 - accuracy: 0.9625 - val_loss: 0.4799 - val_accuracy: 0.9023 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 338/350
283/283 - 3s - loss: 0.2407 - accuracy: 0.9647 - val_loss: 0.4700 - val_accuracy: 0.9037 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 339/350
283/283 - 3s - loss: 0.2452 - accuracy: 0.9597 - val_loss: 0.4669 - val_accuracy: 0.9040 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 340/350
283/283 - 3s - loss: 0.2533 - accuracy: 0.9595 - val_loss: 0.4654 - val_accuracy: 0.9030 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 341/350
283/283 - 3s - loss: 0.2539 - accuracy: 0.9599 - val_loss: 0.4802 - val_accuracy: 0.9047 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 342/350
283/283 - 3s - loss: 0.2416 - accuracy: 0.9634 - val_loss: 0.4625 - val_accuracy: 0.9067 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 343/350
283/283 - 3s - loss: 0.2453 - accuracy: 0.9613 - val_loss: 0.4682 - val_accuracy: 0.9070 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 344/350
283/283 - 3s - loss: 0.2330 - accuracy: 0.9661 - val_loss: 0.4676 - val_accuracy: 0.9067 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 345/350
283/283 - 3s - loss: 0.2338 - accuracy: 0.9641 - val_loss: 0.4593 - val_accuracy: 0.9063 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 346/350
283/283 - 3s - loss: 0.2522 - accuracy: 0.9621 - val_loss: 0.4534 - val_accuracy: 0.9063 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 347/350
283/283 - 3s - loss: 0.2438 - accuracy: 0.9623 - val_loss: 0.4868 - val_accuracy: 0.9000 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 348/350
283/283 - 3s - loss: 0.2538 - accuracy: 0.9610 - val_loss: 0.4613 - val_accuracy: 0.9047 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 349/350
283/283 - 3s - loss: 0.2523 - accuracy: 0.9606 - val_loss: 0.4606 - val_accuracy: 0.9040 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
Epoch 350/350
283/283 - 3s - loss: 0.2479 - accuracy: 0.9613 - val_loss: 0.4845 - val_accuracy: 0.8957 - lr: 1.3422e-04 - 3s/epoch - 9ms/step
94/94 - 0s - loss: 0.4845 - accuracy: 0.8957 - 307ms/epoch - 3ms/step
Baseline Error: 10.43%

df128

	Model	Train Accuracy	Validation Accuracy	Train Loss	Validation Loss	History
0	Base 128 Model	0.985490	0.769333	0.045317	1.295857	<keras.callbacks.History object at 0x7f5a16d74...
1	LeNet	1.000000	0.704000	0.003636	1.203718	<keras.callbacks.History object at 0x7f59ec532...
2	AlexNet	0.105782	0.066667	2.637087	2.787983	<keras.callbacks.History object at 0x7f59ec79e...
3	RegularizedModel	0.964444	0.754333	0.478071	1.542403	<keras.callbacks.History object at 0x7f5942cc5...
4	ExtraThickModel	0.835290	0.766000	0.920840	1.247124	<keras.callbacks.History object at 0x7f59428f2...
5	BNModel	0.737483	0.373000	1.096512	3.277027	<keras.callbacks.History object at 0x7f5960582...
6	BNlrdModel	0.844041	0.747333	0.817036	1.183966	<keras.callbacks.History object at 0x7f594289f...
7	BNlrdRegModel	0.818564	0.852333	0.946453	0.859434	<keras.callbacks.History object at 0x7f5942426...
8	BNlrdRegModelXtraepochs	0.980284	0.932667	0.172018	0.384631	<keras.callbacks.History object at 0x7f59344f9...

We can see that the model has fully converged and has given us an excellent loss and accuracy compared to the other models, we can now save this model and treat it as the best model by far.

model_128BNlrdReg.save('Best128Model.h5')

plt.figure(figsize=(15, 8))

for index, row in df128[['Model', 'History']].iterrows():
    model_name = row['Model']
    history = row['History']
    sns.lineplot(x=range(1, len(history.history['val_accuracy']) + 1), y=history.history['val_accuracy'], label=f'{model_name} Validation')

plt.title('128x128Model Accuracies Over Epochs')
plt.xlabel('Epochs')
plt.ylabel('Accuracy')

plt.legend()

plt.show()

plt.figure(figsize=(15, 8))

for index, row in df128[['Model', 'History']].iterrows():
    model_name = row['Model']
    history = row['History']

    accuracy_diff = [train_acc - val_acc for train_acc, val_acc in zip(history.history['accuracy'], history.history['val_accuracy'])]

    sns.lineplot(x=range(1, len(accuracy_diff) + 1), y=accuracy_diff, label=f'{model_name}')

plt.title('Training Accuracy - Validation Accuracy Difference Over Epochs (128x128)')
plt.xlabel('Epochs')
plt.ylabel('Accuracy Difference (To see overfitting)')

plt.legend()

plt.show()

#convert tets to 128x128 
X_test_resized = tf.image.resize(X_test, (128, 128))
X_test128 = X_test_resized.numpy().reshape(X_test_resized.shape[0], 128, 128, 1).astype('float32') / 255
test_labels = to_categorical(y_test)

#load the saved models weights

model128best = models.load_model('Best128Model.h5')


#evaluate the model on the test set
score = model128best.evaluate(X_test128, test_labels, verbose=0)
print("Test loss:", score[0])
print("Test accuracy:", score[1])

Test loss: 0.46002209186553955
Test accuracy: 0.9110000133514404

128x128 Final Super Model

Same as for the 31x31 model, we will test the different ways to balance the data and see if there are any improvements/differences in the results.

# Convert datasets to NumPy arrays
train_images, train_labels_augmented = zip(*[(image, label) for image, label in train_data.as_numpy_iterator()])

#train set
X_train_aug = np.concatenate(train_images, axis=0)
y_train = np.concatenate(train_labels_augmented, axis=0)

# Resize images to (128, 128) using TensorFlow
X_train_resized = tf.image.resize(X_train_aug, (128, 128))

# Reshape and normalize
X_train128_aug = X_train_resized.numpy().reshape(X_train_resized.shape[0], 128, 128, 1).astype('float32') / 255

X_train128_augmented, y_train_augmented = augmentfill(y_train, augmentation_dict, X_train128_aug)
train_labels_augmented = to_categorical(y_train_augmented)

214
Augmented 214 images out of 214
Category 0: Augmented 214 images out of 214
254
Augmented 254 images out of 254
Category 1: Augmented 254 images out of 254
420
Augmented 420 images out of 535
115
Augmented 535 images out of 535
Category 2: Augmented 535 images out of 535
129
Augmented 129 images out of 129
Category 3: Augmented 129 images out of 129
219
Augmented 219 images out of 219
Category 4: Augmented 219 images out of 219
454
Augmented 454 images out of 454
Category 5: Augmented 454 images out of 454
346
Augmented 346 images out of 609
263
Augmented 609 images out of 609
Category 6: Augmented 609 images out of 609
256
Augmented 256 images out of 699
256
Augmented 512 images out of 699
187
Augmented 699 images out of 699
Category 7: Augmented 699 images out of 699
374
Augmented 374 images out of 374
Category 8: Augmented 374 images out of 374
205
Augmented 205 images out of 205
Category 9: Augmented 205 images out of 205
392
Augmented 392 images out of 392
Category 10: Augmented 392 images out of 392
373
Augmented 373 images out of 582
209
Augmented 582 images out of 582
Category 11: Augmented 582 images out of 582
144
Augmented 144 images out of 144
Category 12: Augmented 144 images out of 144
247
Augmented 247 images out of 708
247
Augmented 494 images out of 708
214
Augmented 708 images out of 708
Category 13: Augmented 708 images out of 708
Category 14: Augmented 0 images out of 0

def Goat128Model():
    Goat128Model = models.Sequential([
        layers.MaxPooling2D((2, 2)),
        layers.Conv2D(64, (3, 3), activation='relu'),
        layers.MaxPooling2D((2, 2)),

        layers.Conv2D(64, (3, 3), activation='relu'),
        layers.Dropout(0.5),

        layers.Flatten(),

        layers.Dense(64, activation='relu', kernel_regularizer=regularizers.l2(0.001)),

        layers.Dropout(0.5),
        layers.Dense(32, activation='relu'),

        layers.BatchNormalization(),
        layers.Dense(15, activation='softmax')
    ])
    return Goat128Model

## Test with just augmented data
model = Goat128Model()

model.compile(optimizer='adam',
                loss= 'categorical_crossentropy',
                metrics=['accuracy'])

history = model.fit(X_train128_augmented, train_labels_augmented, validation_data=(X_val128, validation_labels),
                    epochs=150, batch_size=32, verbose=2, #no class weights
                    callbacks=[ReduceLROnPlateau(monitor='val_loss', factor=0.8, patience=10, verbose=1, mode='auto', min_delta=0.0001, cooldown=0, min_lr=0)])


scores = model.evaluate(X_val128, validation_labels, verbose=2)
print("Baseline Error: %.2f%%" % (100-scores[1]*100))
plot_history(history)
#save the model
model.save('Goat128Model.h5')

Epoch 1/150

2023-11-26 15:32:44.775856: E tensorflow/core/grappler/optimizers/meta_optimizer.cc:954] layout failed: INVALID_ARGUMENT: Size of values 0 does not match size of permutation 4 @ fanin shape insequential_26/dropout_37/dropout/SelectV2-2-TransposeNHWCToNCHW-LayoutOptimizer

448/448 - 5s - loss: 2.6684 - accuracy: 0.1368 - val_loss: 2.4384 - val_accuracy: 0.3227 - lr: 0.0010 - 5s/epoch - 12ms/step
Epoch 2/150
448/448 - 2s - loss: 2.3791 - accuracy: 0.2314 - val_loss: 2.3074 - val_accuracy: 0.3077 - lr: 0.0010 - 2s/epoch - 5ms/step
Epoch 3/150
448/448 - 2s - loss: 2.2801 - accuracy: 0.2824 - val_loss: 2.2174 - val_accuracy: 0.3433 - lr: 0.0010 - 2s/epoch - 5ms/step
Epoch 4/150
448/448 - 2s - loss: 2.2408 - accuracy: 0.3180 - val_loss: 1.8148 - val_accuracy: 0.5583 - lr: 0.0010 - 2s/epoch - 5ms/step
Epoch 5/150
448/448 - 2s - loss: 2.1665 - accuracy: 0.3541 - val_loss: 1.8193 - val_accuracy: 0.5200 - lr: 0.0010 - 2s/epoch - 5ms/step
Epoch 6/150
448/448 - 2s - loss: 2.1445 - accuracy: 0.3696 - val_loss: 1.8455 - val_accuracy: 0.5717 - lr: 0.0010 - 2s/epoch - 5ms/step
Epoch 7/150
448/448 - 2s - loss: 2.0904 - accuracy: 0.4020 - val_loss: 1.7306 - val_accuracy: 0.6083 - lr: 0.0010 - 2s/epoch - 6ms/step
Epoch 8/150
448/448 - 2s - loss: 2.0414 - accuracy: 0.4202 - val_loss: 1.7930 - val_accuracy: 0.5527 - lr: 0.0010 - 2s/epoch - 5ms/step
Epoch 9/150
448/448 - 2s - loss: 2.0364 - accuracy: 0.4255 - val_loss: 1.6595 - val_accuracy: 0.6233 - lr: 0.0010 - 2s/epoch - 5ms/step
Epoch 10/150
448/448 - 2s - loss: 2.0405 - accuracy: 0.4385 - val_loss: 1.6953 - val_accuracy: 0.6407 - lr: 0.0010 - 2s/epoch - 5ms/step
Epoch 11/150
448/448 - 2s - loss: 2.0001 - accuracy: 0.4565 - val_loss: 1.6250 - val_accuracy: 0.6377 - lr: 0.0010 - 2s/epoch - 5ms/step
Epoch 12/150
448/448 - 2s - loss: 2.0130 - accuracy: 0.4562 - val_loss: 1.7206 - val_accuracy: 0.5903 - lr: 0.0010 - 2s/epoch - 5ms/step
Epoch 13/150
448/448 - 2s - loss: 1.9791 - accuracy: 0.4722 - val_loss: 1.6351 - val_accuracy: 0.6400 - lr: 0.0010 - 2s/epoch - 5ms/step
Epoch 14/150
448/448 - 2s - loss: 1.9626 - accuracy: 0.4786 - val_loss: 1.5224 - val_accuracy: 0.6727 - lr: 0.0010 - 2s/epoch - 5ms/step
Epoch 15/150
448/448 - 2s - loss: 1.9537 - accuracy: 0.4789 - val_loss: 1.5791 - val_accuracy: 0.6670 - lr: 0.0010 - 2s/epoch - 6ms/step
Epoch 16/150
448/448 - 2s - loss: 1.9524 - accuracy: 0.4853 - val_loss: 1.5554 - val_accuracy: 0.6837 - lr: 0.0010 - 2s/epoch - 5ms/step
Epoch 17/150
448/448 - 2s - loss: 1.9222 - accuracy: 0.5024 - val_loss: 1.7579 - val_accuracy: 0.6107 - lr: 0.0010 - 2s/epoch - 5ms/step
Epoch 18/150
448/448 - 2s - loss: 1.9249 - accuracy: 0.5021 - val_loss: 1.6826 - val_accuracy: 0.6050 - lr: 0.0010 - 2s/epoch - 5ms/step
Epoch 19/150
448/448 - 2s - loss: 1.9233 - accuracy: 0.5073 - val_loss: 1.6363 - val_accuracy: 0.6450 - lr: 0.0010 - 2s/epoch - 5ms/step
Epoch 20/150
448/448 - 2s - loss: 1.9027 - accuracy: 0.5088 - val_loss: 1.6191 - val_accuracy: 0.6533 - lr: 0.0010 - 2s/epoch - 5ms/step
Epoch 21/150
448/448 - 2s - loss: 1.8853 - accuracy: 0.5178 - val_loss: 1.5389 - val_accuracy: 0.6860 - lr: 0.0010 - 2s/epoch - 5ms/step
Epoch 22/150
448/448 - 2s - loss: 1.8918 - accuracy: 0.5201 - val_loss: 1.6431 - val_accuracy: 0.6440 - lr: 0.0010 - 2s/epoch - 5ms/step
Epoch 23/150
448/448 - 2s - loss: 1.8911 - accuracy: 0.5237 - val_loss: 1.7688 - val_accuracy: 0.5983 - lr: 0.0010 - 2s/epoch - 5ms/step
Epoch 24/150

Epoch 24: ReduceLROnPlateau reducing learning rate to 0.000800000037997961.
448/448 - 2s - loss: 1.9000 - accuracy: 0.5193 - val_loss: 1.5294 - val_accuracy: 0.7000 - lr: 0.0010 - 2s/epoch - 5ms/step
Epoch 25/150
448/448 - 2s - loss: 1.8225 - accuracy: 0.5409 - val_loss: 1.5313 - val_accuracy: 0.6720 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 26/150
448/448 - 2s - loss: 1.8043 - accuracy: 0.5407 - val_loss: 1.4996 - val_accuracy: 0.6823 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 27/150
448/448 - 2s - loss: 1.7793 - accuracy: 0.5462 - val_loss: 1.4674 - val_accuracy: 0.6777 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 28/150
448/448 - 2s - loss: 1.7507 - accuracy: 0.5543 - val_loss: 1.5221 - val_accuracy: 0.6610 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 29/150
448/448 - 2s - loss: 1.7416 - accuracy: 0.5611 - val_loss: 1.5024 - val_accuracy: 0.6583 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 30/150
448/448 - 2s - loss: 1.7626 - accuracy: 0.5502 - val_loss: 1.4397 - val_accuracy: 0.6967 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 31/150
448/448 - 2s - loss: 1.7491 - accuracy: 0.5548 - val_loss: 1.4302 - val_accuracy: 0.6977 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 32/150
448/448 - 2s - loss: 1.7289 - accuracy: 0.5661 - val_loss: 1.4866 - val_accuracy: 0.6557 - lr: 8.0000e-04 - 2s/epoch - 6ms/step
Epoch 33/150
448/448 - 2s - loss: 1.7621 - accuracy: 0.5586 - val_loss: 1.4604 - val_accuracy: 0.6833 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 34/150
448/448 - 2s - loss: 1.7042 - accuracy: 0.5749 - val_loss: 1.4293 - val_accuracy: 0.6860 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 35/150
448/448 - 2s - loss: 1.7112 - accuracy: 0.5733 - val_loss: 1.5784 - val_accuracy: 0.6357 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 36/150
448/448 - 2s - loss: 1.7050 - accuracy: 0.5745 - val_loss: 1.5775 - val_accuracy: 0.6447 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 37/150
448/448 - 2s - loss: 1.6928 - accuracy: 0.5790 - val_loss: 1.4507 - val_accuracy: 0.6757 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 38/150
448/448 - 2s - loss: 1.7021 - accuracy: 0.5782 - val_loss: 1.6419 - val_accuracy: 0.6213 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 39/150
448/448 - 2s - loss: 1.6823 - accuracy: 0.5826 - val_loss: 1.4355 - val_accuracy: 0.6883 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 40/150
448/448 - 2s - loss: 1.6654 - accuracy: 0.5869 - val_loss: 1.4459 - val_accuracy: 0.6803 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 41/150
448/448 - 2s - loss: 1.7178 - accuracy: 0.5727 - val_loss: 1.4524 - val_accuracy: 0.6700 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 42/150
448/448 - 2s - loss: 1.6905 - accuracy: 0.5835 - val_loss: 1.4272 - val_accuracy: 0.7030 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 43/150
448/448 - 2s - loss: 1.6962 - accuracy: 0.5772 - val_loss: 1.4246 - val_accuracy: 0.6963 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 44/150
448/448 - 2s - loss: 1.6977 - accuracy: 0.5822 - val_loss: 1.5071 - val_accuracy: 0.6513 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 45/150
448/448 - 2s - loss: 1.6877 - accuracy: 0.5814 - val_loss: 1.5003 - val_accuracy: 0.6530 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 46/150
448/448 - 2s - loss: 1.7013 - accuracy: 0.5816 - val_loss: 1.4574 - val_accuracy: 0.6747 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 47/150
448/448 - 2s - loss: 1.6604 - accuracy: 0.5948 - val_loss: 1.3953 - val_accuracy: 0.6920 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 48/150
448/448 - 2s - loss: 1.6661 - accuracy: 0.5903 - val_loss: 1.4611 - val_accuracy: 0.6837 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 49/150
448/448 - 2s - loss: 1.6947 - accuracy: 0.5807 - val_loss: 1.4316 - val_accuracy: 0.6943 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 50/150
448/448 - 2s - loss: 1.6588 - accuracy: 0.5951 - val_loss: 1.6163 - val_accuracy: 0.6117 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 51/150
448/448 - 2s - loss: 1.6511 - accuracy: 0.5952 - val_loss: 1.4198 - val_accuracy: 0.6910 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 52/150
448/448 - 2s - loss: 1.6461 - accuracy: 0.6015 - val_loss: 1.4608 - val_accuracy: 0.6510 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 53/150
448/448 - 2s - loss: 1.6580 - accuracy: 0.5952 - val_loss: 1.4299 - val_accuracy: 0.6800 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 54/150
448/448 - 2s - loss: 1.6552 - accuracy: 0.5948 - val_loss: 1.4900 - val_accuracy: 0.6637 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 55/150
448/448 - 2s - loss: 1.6556 - accuracy: 0.5980 - val_loss: 1.4649 - val_accuracy: 0.6717 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 56/150
448/448 - 2s - loss: 1.6422 - accuracy: 0.6025 - val_loss: 1.5799 - val_accuracy: 0.6450 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 57/150

Epoch 57: ReduceLROnPlateau reducing learning rate to 0.0006400000303983689.
448/448 - 2s - loss: 1.6452 - accuracy: 0.5992 - val_loss: 1.6388 - val_accuracy: 0.6107 - lr: 8.0000e-04 - 2s/epoch - 5ms/step
Epoch 58/150
448/448 - 2s - loss: 1.5925 - accuracy: 0.6139 - val_loss: 1.5221 - val_accuracy: 0.6460 - lr: 6.4000e-04 - 2s/epoch - 6ms/step
Epoch 59/150
448/448 - 2s - loss: 1.5692 - accuracy: 0.6103 - val_loss: 1.6728 - val_accuracy: 0.5933 - lr: 6.4000e-04 - 2s/epoch - 5ms/step
Epoch 60/150
448/448 - 2s - loss: 1.5443 - accuracy: 0.6195 - val_loss: 1.3427 - val_accuracy: 0.7030 - lr: 6.4000e-04 - 2s/epoch - 5ms/step
Epoch 61/150
448/448 - 2s - loss: 1.5185 - accuracy: 0.6272 - val_loss: 1.4079 - val_accuracy: 0.6883 - lr: 6.4000e-04 - 2s/epoch - 5ms/step
Epoch 62/150
448/448 - 2s - loss: 1.5056 - accuracy: 0.6276 - val_loss: 1.3728 - val_accuracy: 0.6960 - lr: 6.4000e-04 - 2s/epoch - 5ms/step
Epoch 63/150
448/448 - 2s - loss: 1.5053 - accuracy: 0.6279 - val_loss: 1.4205 - val_accuracy: 0.6787 - lr: 6.4000e-04 - 2s/epoch - 5ms/step
Epoch 64/150
448/448 - 2s - loss: 1.5027 - accuracy: 0.6299 - val_loss: 1.4080 - val_accuracy: 0.6903 - lr: 6.4000e-04 - 2s/epoch - 5ms/step
Epoch 65/150
448/448 - 2s - loss: 1.4913 - accuracy: 0.6335 - val_loss: 1.4786 - val_accuracy: 0.6480 - lr: 6.4000e-04 - 2s/epoch - 5ms/step
Epoch 66/150
448/448 - 2s - loss: 1.4905 - accuracy: 0.6286 - val_loss: 1.3419 - val_accuracy: 0.6983 - lr: 6.4000e-04 - 2s/epoch - 5ms/step
Epoch 67/150
448/448 - 2s - loss: 1.4881 - accuracy: 0.6279 - val_loss: 1.3812 - val_accuracy: 0.6777 - lr: 6.4000e-04 - 2s/epoch - 5ms/step
Epoch 68/150
448/448 - 2s - loss: 1.5087 - accuracy: 0.6242 - val_loss: 1.4302 - val_accuracy: 0.6607 - lr: 6.4000e-04 - 2s/epoch - 5ms/step
Epoch 69/150
448/448 - 2s - loss: 1.5181 - accuracy: 0.6242 - val_loss: 1.3512 - val_accuracy: 0.6980 - lr: 6.4000e-04 - 2s/epoch - 5ms/step
Epoch 70/150
448/448 - 2s - loss: 1.4804 - accuracy: 0.6353 - val_loss: 1.4969 - val_accuracy: 0.6483 - lr: 6.4000e-04 - 2s/epoch - 5ms/step
Epoch 71/150
448/448 - 2s - loss: 1.5015 - accuracy: 0.6286 - val_loss: 1.3731 - val_accuracy: 0.6857 - lr: 6.4000e-04 - 2s/epoch - 5ms/step
Epoch 72/150
448/448 - 2s - loss: 1.4792 - accuracy: 0.6321 - val_loss: 1.3833 - val_accuracy: 0.6940 - lr: 6.4000e-04 - 2s/epoch - 5ms/step
Epoch 73/150
448/448 - 2s - loss: 1.4745 - accuracy: 0.6395 - val_loss: 1.3796 - val_accuracy: 0.6920 - lr: 6.4000e-04 - 2s/epoch - 5ms/step
Epoch 74/150
448/448 - 2s - loss: 1.4649 - accuracy: 0.6441 - val_loss: 1.3028 - val_accuracy: 0.7103 - lr: 6.4000e-04 - 2s/epoch - 5ms/step
Epoch 75/150
448/448 - 2s - loss: 1.4788 - accuracy: 0.6365 - val_loss: 1.5579 - val_accuracy: 0.6233 - lr: 6.4000e-04 - 2s/epoch - 5ms/step
Epoch 76/150
448/448 - 2s - loss: 1.4832 - accuracy: 0.6336 - val_loss: 1.3890 - val_accuracy: 0.6843 - lr: 6.4000e-04 - 2s/epoch - 5ms/step
Epoch 77/150
448/448 - 2s - loss: 1.4406 - accuracy: 0.6428 - val_loss: 1.4956 - val_accuracy: 0.6497 - lr: 6.4000e-04 - 2s/epoch - 5ms/step
Epoch 78/150
448/448 - 2s - loss: 1.4773 - accuracy: 0.6329 - val_loss: 1.4744 - val_accuracy: 0.6633 - lr: 6.4000e-04 - 2s/epoch - 5ms/step
Epoch 79/150
448/448 - 2s - loss: 1.4700 - accuracy: 0.6335 - val_loss: 1.4001 - val_accuracy: 0.6813 - lr: 6.4000e-04 - 2s/epoch - 5ms/step
Epoch 80/150
448/448 - 2s - loss: 1.4711 - accuracy: 0.6413 - val_loss: 1.3289 - val_accuracy: 0.7227 - lr: 6.4000e-04 - 2s/epoch - 5ms/step
Epoch 81/150

model = models.load_model('Goat128Model.h5')

#evaluate the model on the test set
score = model.evaluate(X_test128, test_labels, verbose=0)
print("Test loss:", score[0])
print("Test accuracy:", score[1])

Final 128x128 Evaluation

We can see that the model trained on the augmented data has poor performance compared to our original model. Although, one thing to note that is taht the model converges faster on the augmented data, although at a lower accuracy. To solve this issue, we can consider tuning the augmentation function to augment the data in another way. However, as tuning is not a part of the assingment, i will not be doing so.

The final model for 128x128 would be the model trained on the original data with class weights, as it has the highest test accuracy and lowest test loss out of all the models.